MNSIT(Vector_Victors).ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "O3taMTaQi35p"
   },
   "source": [
    "# MNIST - Digit Recognizer"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "oucgPfati35r"
   },
   "source": [
    "Source of the dataset: https://www.kaggle.com/competitions/digit-recognizer/data\n",
    "\n",
    "We are using SVM as one of the classifications methods to identify handwritten digits based on the pixels values in the data set. More information on data sets is in the next section:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "bbI0wakBi35s"
   },
   "source": [
    "### Explaination of data sets"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "k96BH50Bi35s"
   },
   "source": [
    "The data files train.csv and test.csv contain gray-scale images of hand-drawn digits, from zero through nine.\n",
    "\n",
    "Each image is 28 pixels in height and 28 pixels in width, for a total of 784 pixels in total. Each pixel has a single pixel-value associated with it, indicating the lightness or darkness of that pixel, with higher numbers meaning darker. This pixel-value is an integer between 0 and 255, inclusive.\n",
    "\n",
    "The training data set, (train.csv), has 785 columns. The first column, called \"label\", is the digit that was drawn by the user. The rest of the columns contain the pixel-values of the associated image.\n",
    "\n",
    "Each pixel column in the training set has a name like pixelx, where x is an integer between 0 and 783, inclusive. To locate this pixel on the image, suppose that we have decomposed x as x = i * 28 + j, where i and j are integers between 0 and 27, inclusive. Then pixelx is located on row i and column j of a 28 x 28 matrix, (indexing by zero)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Classification Methods"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here, we are trying to use two machine learning methods - **Support Vector Machines** and **Convolutional Neural Network**. We will also try to figure out the optimal parameters to get the best accuracies"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SVM"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We are using SVM as one of the classifications methods to identify handwritten digits based on the pixel’s values in the data set. SVM is highly effective in high dimensional spaces. This uses a subset of training points called support vectors in the decision functions making it more memory efficient. \n",
    "In our problem, we used 2 models, Linear and RBF models of SVM, evaluate the performances of each model and choose the high performing model. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "z5lGHb22i35s"
   },
   "source": [
    "## Importing Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "id": "ByT9sTZMi35s"
   },
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "id": "qDEaVvLAi35t"
   },
   "outputs": [],
   "source": [
    "'''Importing all the necessary libraries for SVM'''\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "# For Visualisation\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "# For Model building\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# For metric evaluation\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "# For Cross validation and regularization model building\n",
    "from sklearn.model_selection import KFold\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "# For scaling\n",
    "from sklearn.preprocessing import scale"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "YSOVZKc7i35u"
   },
   "source": [
    "## Data Reading"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 300
    },
    "id": "ARvQmSvXi35u",
    "outputId": "ba352c76-db31-4f20-f777-5236c8f7a1ad"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>label</th>\n",
       "      <th>pixel0</th>\n",
       "      <th>pixel1</th>\n",
       "      <th>pixel2</th>\n",
       "      <th>pixel3</th>\n",
       "      <th>pixel4</th>\n",
       "      <th>pixel5</th>\n",
       "      <th>pixel6</th>\n",
       "      <th>pixel7</th>\n",
       "      <th>pixel8</th>\n",
       "      <th>...</th>\n",
       "      <th>pixel774</th>\n",
       "      <th>pixel775</th>\n",
       "      <th>pixel776</th>\n",
       "      <th>pixel777</th>\n",
       "      <th>pixel778</th>\n",
       "      <th>pixel779</th>\n",
       "      <th>pixel780</th>\n",
       "      <th>pixel781</th>\n",
       "      <th>pixel782</th>\n",
       "      <th>pixel783</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 785 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   label  pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  \\\n",
       "0      1       0       0       0       0       0       0       0       0   \n",
       "1      0       0       0       0       0       0       0       0       0   \n",
       "2      1       0       0       0       0       0       0       0       0   \n",
       "3      4       0       0       0       0       0       0       0       0   \n",
       "4      0       0       0       0       0       0       0       0       0   \n",
       "\n",
       "   pixel8  ...  pixel774  pixel775  pixel776  pixel777  pixel778  pixel779  \\\n",
       "0       0  ...         0         0         0         0         0         0   \n",
       "1       0  ...         0         0         0         0         0         0   \n",
       "2       0  ...         0         0         0         0         0         0   \n",
       "3       0  ...         0         0         0         0         0         0   \n",
       "4       0  ...         0         0         0         0         0         0   \n",
       "\n",
       "   pixel780  pixel781  pixel782  pixel783  \n",
       "0         0         0         0         0  \n",
       "1         0         0         0         0  \n",
       "2         0         0         0         0  \n",
       "3         0         0         0         0  \n",
       "4         0         0         0         0  \n",
       "\n",
       "[5 rows x 785 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''Downloaded the data set from Kaggle: https://www.kaggle.com/competitions/digit-recognizer/data. Extracting only the \n",
    "train data set '''\n",
    "digit = pd.read_csv(\"train.csv\")\n",
    "digit.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "s68joaqEi35v"
   },
   "source": [
    "The training data set, (train.csv), has 785 columns. The first column is \"label\" - indicates the number drawn based on the pixel values in the subsequent columns(the rest of 784 columns)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "xdiJW1bGi35v",
    "outputId": "e4b28fa9-4c78-486f-ff65-08458bdbc76e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(42000, 785)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "digit.shape #Each image is 28 pixels in height and 28 pixels in width, for a total of 784 pixels in total = 784 columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "mVFCQ0fUi35v",
    "outputId": "d4a9167d-f470-4336-a3a1-8a6761f7e2ca"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['label', 'pixel0', 'pixel1', 'pixel2', 'pixel3', 'pixel4', 'pixel5',\n",
       "       'pixel6', 'pixel7', 'pixel8',\n",
       "       ...\n",
       "       'pixel774', 'pixel775', 'pixel776', 'pixel777', 'pixel778', 'pixel779',\n",
       "       'pixel780', 'pixel781', 'pixel782', 'pixel783'],\n",
       "      dtype='object', length=785)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''there are 785 columns with 784 as pixel cols and one of cols as 'label' '''\n",
    "digit.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "WFfNt942i35w",
    "outputId": "40d65cc0-d38c-41e9-d5a5-d2c112e962c4"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    4132\n",
       "1    4684\n",
       "2    4177\n",
       "3    4351\n",
       "4    4072\n",
       "5    3795\n",
       "6    4137\n",
       "7    4401\n",
       "8    4063\n",
       "9    4188\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''Checking for amount of times the value has been drawn'''\n",
    "\n",
    "digit.label.value_counts().sort_index(ascending=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 279
    },
    "id": "_eS3p3c-i35w",
    "outputId": "078ca132-fa22-42a6-c928-66c2efbcb803"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#representing the counts on a bar plot\n",
    "sns.countplot(digit[\"label\"])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 394
    },
    "id": "12godpV6i35w",
    "outputId": "392b79fa-3c7c-4421-fbe2-37e10d26e0a8"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>label</th>\n",
       "      <th>pixel0</th>\n",
       "      <th>pixel1</th>\n",
       "      <th>pixel2</th>\n",
       "      <th>pixel3</th>\n",
       "      <th>pixel4</th>\n",
       "      <th>pixel5</th>\n",
       "      <th>pixel6</th>\n",
       "      <th>pixel7</th>\n",
       "      <th>pixel8</th>\n",
       "      <th>...</th>\n",
       "      <th>pixel774</th>\n",
       "      <th>pixel775</th>\n",
       "      <th>pixel776</th>\n",
       "      <th>pixel777</th>\n",
       "      <th>pixel778</th>\n",
       "      <th>pixel779</th>\n",
       "      <th>pixel780</th>\n",
       "      <th>pixel781</th>\n",
       "      <th>pixel782</th>\n",
       "      <th>pixel783</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>42000.000000</td>\n",
       "      <td>42000.0</td>\n",
       "      <td>42000.0</td>\n",
       "      <td>42000.0</td>\n",
       "      <td>42000.0</td>\n",
       "      <td>42000.0</td>\n",
       "      <td>42000.0</td>\n",
       "      <td>42000.0</td>\n",
       "      <td>42000.0</td>\n",
       "      <td>42000.0</td>\n",
       "      <td>...</td>\n",
       "      <td>42000.000000</td>\n",
       "      <td>42000.000000</td>\n",
       "      <td>42000.000000</td>\n",
       "      <td>42000.00000</td>\n",
       "      <td>42000.000000</td>\n",
       "      <td>42000.000000</td>\n",
       "      <td>42000.0</td>\n",
       "      <td>42000.0</td>\n",
       "      <td>42000.0</td>\n",
       "      <td>42000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>4.456643</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.219286</td>\n",
       "      <td>0.117095</td>\n",
       "      <td>0.059024</td>\n",
       "      <td>0.02019</td>\n",
       "      <td>0.017238</td>\n",
       "      <td>0.002857</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>2.887730</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>6.312890</td>\n",
       "      <td>4.633819</td>\n",
       "      <td>3.274488</td>\n",
       "      <td>1.75987</td>\n",
       "      <td>1.894498</td>\n",
       "      <td>0.414264</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>2.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>4.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>7.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>9.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>254.000000</td>\n",
       "      <td>254.000000</td>\n",
       "      <td>253.000000</td>\n",
       "      <td>253.00000</td>\n",
       "      <td>254.000000</td>\n",
       "      <td>62.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>8 rows × 785 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              label   pixel0   pixel1   pixel2   pixel3   pixel4   pixel5  \\\n",
       "count  42000.000000  42000.0  42000.0  42000.0  42000.0  42000.0  42000.0   \n",
       "mean       4.456643      0.0      0.0      0.0      0.0      0.0      0.0   \n",
       "std        2.887730      0.0      0.0      0.0      0.0      0.0      0.0   \n",
       "min        0.000000      0.0      0.0      0.0      0.0      0.0      0.0   \n",
       "25%        2.000000      0.0      0.0      0.0      0.0      0.0      0.0   \n",
       "50%        4.000000      0.0      0.0      0.0      0.0      0.0      0.0   \n",
       "75%        7.000000      0.0      0.0      0.0      0.0      0.0      0.0   \n",
       "max        9.000000      0.0      0.0      0.0      0.0      0.0      0.0   \n",
       "\n",
       "        pixel6   pixel7   pixel8  ...      pixel774      pixel775  \\\n",
       "count  42000.0  42000.0  42000.0  ...  42000.000000  42000.000000   \n",
       "mean       0.0      0.0      0.0  ...      0.219286      0.117095   \n",
       "std        0.0      0.0      0.0  ...      6.312890      4.633819   \n",
       "min        0.0      0.0      0.0  ...      0.000000      0.000000   \n",
       "25%        0.0      0.0      0.0  ...      0.000000      0.000000   \n",
       "50%        0.0      0.0      0.0  ...      0.000000      0.000000   \n",
       "75%        0.0      0.0      0.0  ...      0.000000      0.000000   \n",
       "max        0.0      0.0      0.0  ...    254.000000    254.000000   \n",
       "\n",
       "           pixel776     pixel777      pixel778      pixel779  pixel780  \\\n",
       "count  42000.000000  42000.00000  42000.000000  42000.000000   42000.0   \n",
       "mean       0.059024      0.02019      0.017238      0.002857       0.0   \n",
       "std        3.274488      1.75987      1.894498      0.414264       0.0   \n",
       "min        0.000000      0.00000      0.000000      0.000000       0.0   \n",
       "25%        0.000000      0.00000      0.000000      0.000000       0.0   \n",
       "50%        0.000000      0.00000      0.000000      0.000000       0.0   \n",
       "75%        0.000000      0.00000      0.000000      0.000000       0.0   \n",
       "max      253.000000    253.00000    254.000000     62.000000       0.0   \n",
       "\n",
       "       pixel781  pixel782  pixel783  \n",
       "count   42000.0   42000.0   42000.0  \n",
       "mean        0.0       0.0       0.0  \n",
       "std         0.0       0.0       0.0  \n",
       "min         0.0       0.0       0.0  \n",
       "25%         0.0       0.0       0.0  \n",
       "50%         0.0       0.0       0.0  \n",
       "75%         0.0       0.0       0.0  \n",
       "max         0.0       0.0       0.0  \n",
       "\n",
       "[8 rows x 785 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "digit.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "x0_qwq8Oi35w"
   },
   "source": [
    "### Data Cleaning and Scaling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "zDc2vx3Ii35w",
    "outputId": "db59d454-1851-484d-f6f1-b20c158a083a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''Checking for missing values in the train data set '''\n",
    "\n",
    "digit.isnull().sort_index(axis =0).values.any()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "CBYL8WeEi35x"
   },
   "source": [
    "\"False\" indicates that there are no null/missing values in the dataset. We have a clean dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 488
    },
    "id": "dF1vBNbEi35x",
    "outputId": "eb76db88-4912-4fde-d467-fb0ba4ddfd3e"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>pixel0</th>\n",
       "      <th>pixel1</th>\n",
       "      <th>pixel2</th>\n",
       "      <th>pixel3</th>\n",
       "      <th>pixel4</th>\n",
       "      <th>pixel5</th>\n",
       "      <th>pixel6</th>\n",
       "      <th>pixel7</th>\n",
       "      <th>pixel8</th>\n",
       "      <th>pixel9</th>\n",
       "      <th>...</th>\n",
       "      <th>pixel774</th>\n",
       "      <th>pixel775</th>\n",
       "      <th>pixel776</th>\n",
       "      <th>pixel777</th>\n",
       "      <th>pixel778</th>\n",
       "      <th>pixel779</th>\n",
       "      <th>pixel780</th>\n",
       "      <th>pixel781</th>\n",
       "      <th>pixel782</th>\n",
       "      <th>pixel783</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>label</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>1.846626</td>\n",
       "      <td>0.939105</td>\n",
       "      <td>0.397182</td>\n",
       "      <td>0.086344</td>\n",
       "      <td>0.118155</td>\n",
       "      <td>0.027267</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.258596</td>\n",
       "      <td>0.187440</td>\n",
       "      <td>0.174546</td>\n",
       "      <td>0.111748</td>\n",
       "      <td>0.048711</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10 rows × 784 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  pixel8  \\\n",
       "label                                                                           \n",
       "0         0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0   \n",
       "1         0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0   \n",
       "2         0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0   \n",
       "3         0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0   \n",
       "4         0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0   \n",
       "5         0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0   \n",
       "6         0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0   \n",
       "7         0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0   \n",
       "8         0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0   \n",
       "9         0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0     0.0   \n",
       "\n",
       "       pixel9  ...  pixel774  pixel775  pixel776  pixel777  pixel778  \\\n",
       "label          ...                                                     \n",
       "0         0.0  ...  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
       "1         0.0  ...  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
       "2         0.0  ...  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
       "3         0.0  ...  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
       "4         0.0  ...  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
       "5         0.0  ...  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
       "6         0.0  ...  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
       "7         0.0  ...  1.846626  0.939105  0.397182  0.086344  0.118155   \n",
       "8         0.0  ...  0.000000  0.000000  0.000000  0.000000  0.000000   \n",
       "9         0.0  ...  0.258596  0.187440  0.174546  0.111748  0.048711   \n",
       "\n",
       "       pixel779  pixel780  pixel781  pixel782  pixel783  \n",
       "label                                                    \n",
       "0      0.000000       0.0       0.0       0.0       0.0  \n",
       "1      0.000000       0.0       0.0       0.0       0.0  \n",
       "2      0.000000       0.0       0.0       0.0       0.0  \n",
       "3      0.000000       0.0       0.0       0.0       0.0  \n",
       "4      0.000000       0.0       0.0       0.0       0.0  \n",
       "5      0.000000       0.0       0.0       0.0       0.0  \n",
       "6      0.000000       0.0       0.0       0.0       0.0  \n",
       "7      0.027267       0.0       0.0       0.0       0.0  \n",
       "8      0.000000       0.0       0.0       0.0       0.0  \n",
       "9      0.000000       0.0       0.0       0.0       0.0  \n",
       "\n",
       "[10 rows x 784 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''Checking for correlation between various labels by taking their means'''\n",
    "mean_digit = digit.groupby('label').mean()\n",
    "mean_digit.sort_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 914
    },
    "id": "yabG9lVsi35x",
    "outputId": "ac121495-a55b-4bf5-ee2b-908849c6067d"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:ylabel='label'>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1080 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(20,15))\n",
    "sns.heatmap(mean_digit)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "po6Ovd8Vi35y",
    "outputId": "8af28454-331f-4646-990b-0f8f81503ef9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 42000 entries, 0 to 41999\n",
      "Columns: 785 entries, label to pixel783\n",
      "dtypes: int64(785)\n",
      "memory usage: 251.5 MB\n"
     ]
    }
   ],
   "source": [
    "'''The whole dataset is of numerical datatype. There are NO categorical variables. There is no need to convert them\n",
    "categorical datatype'''\n",
    "digit.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 300
    },
    "id": "Xyyj-Xcui35y",
    "outputId": "6b20087d-ab95-442f-ea95-98e0ec5bf482"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>label</th>\n",
       "      <th>pixel0</th>\n",
       "      <th>pixel1</th>\n",
       "      <th>pixel2</th>\n",
       "      <th>pixel3</th>\n",
       "      <th>pixel4</th>\n",
       "      <th>pixel5</th>\n",
       "      <th>pixel6</th>\n",
       "      <th>pixel7</th>\n",
       "      <th>pixel8</th>\n",
       "      <th>...</th>\n",
       "      <th>pixel774</th>\n",
       "      <th>pixel775</th>\n",
       "      <th>pixel776</th>\n",
       "      <th>pixel777</th>\n",
       "      <th>pixel778</th>\n",
       "      <th>pixel779</th>\n",
       "      <th>pixel780</th>\n",
       "      <th>pixel781</th>\n",
       "      <th>pixel782</th>\n",
       "      <th>pixel783</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 785 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   label  pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  \\\n",
       "0      1       0       0       0       0       0       0       0       0   \n",
       "1      0       0       0       0       0       0       0       0       0   \n",
       "2      1       0       0       0       0       0       0       0       0   \n",
       "3      4       0       0       0       0       0       0       0       0   \n",
       "4      0       0       0       0       0       0       0       0       0   \n",
       "\n",
       "   pixel8  ...  pixel774  pixel775  pixel776  pixel777  pixel778  pixel779  \\\n",
       "0       0  ...         0         0         0         0         0         0   \n",
       "1       0  ...         0         0         0         0         0         0   \n",
       "2       0  ...         0         0         0         0         0         0   \n",
       "3       0  ...         0         0         0         0         0         0   \n",
       "4       0  ...         0         0         0         0         0         0   \n",
       "\n",
       "   pixel780  pixel781  pixel782  pixel783  \n",
       "0         0         0         0         0  \n",
       "1         0         0         0         0  \n",
       "2         0         0         0         0  \n",
       "3         0         0         0         0  \n",
       "4         0         0         0         0  \n",
       "\n",
       "[5 rows x 785 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#digit = digit.sort_values('label')\n",
    "digit.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 283
    },
    "id": "GM64AgD9i35y",
    "outputId": "8c37b764-cfbb-441b-8abf-933f2adb6a25"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Digit Label: 4\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "'''Reading one of the digits'''\n",
    "\n",
    "digit_arr = digit.loc[3, \"pixel0\":]\n",
    "arr = np.array(digit_arr) \n",
    "\n",
    "image_arr = np.reshape(arr, (28,28))\n",
    "\n",
    "digit_image = plt.imshow(image_arr, cmap=plt.cm.binary)\n",
    "plt.colorbar(digit_image)\n",
    "print(\"Digit Label: {}\".format(digit.loc[3, \"label\"]))\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "byLzS9hei35z"
   },
   "source": [
    "**Few Conclusions based on EDA:**\n",
    "- No need of any outlier treatment as we have to consider all the values. All these values contribute to the construction of a pixelated image.\n",
    "- No need to use any dummy variables or convert them to categorical data type as the whole data set is of numerical or integer data type. \n",
    "- Scaling and Normalization can be performed. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "id": "ZVg7h3Uhi35z"
   },
   "outputs": [],
   "source": [
    "'''Spliting the label column from the dataset'''\n",
    "X = digit.drop(\"label\", axis = 1)\n",
    "y = digit['label']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "T2cAjwWli35z"
   },
   "source": [
    "### Train-Valid split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "Q1wfOSAii35z"
   },
   "outputs": [],
   "source": [
    "'''Spliting the dataset into Train and Valid. Alloting 30% of the data to validation data set'''\n",
    "X_train, X_valid, y_train, y_valid = train_test_split(X, y, stratify = y, test_size = 0.2, random_state = 101)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 0
    },
    "id": "PQb6K9E2i35z",
    "outputId": "b3d404f7-ab53-4092-9803-057fd216a0f6"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x864 with 30 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "'''Plotting some digits from train dataset'''\n",
    "\n",
    "plt.figure(figsize=(14,12))\n",
    "for digit_num in range(0,30):\n",
    "    plt.subplot(7,10,digit_num+1)\n",
    "    grid_data = X_train.iloc[digit_num].values.reshape(28,28)  # reshape from 1d to 2d pixel array\n",
    "    plt.imshow(grid_data, interpolation = \"none\", cmap = \"afmhot\")\n",
    "    plt.xticks([])\n",
    "    plt.yticks([])\n",
    "plt.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "vRJ1fTTTi350",
    "outputId": "62132910-ec03-45be-88aa-664932930d89"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train_pixels =  (33600, 784)\n",
      "Train_o_label =  (33600,)\n",
      "Valid_pixels =  (8400, 784)\n",
      "Valid_o_label =  (8400,)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "'''Dimensions of Train and Validation datasets'''\n",
    "print(\"Train_pixels = \", format(X_train.shape) + \"\\n\"\n",
    "     \"Train_o_label = \", format(y_train.shape) + \"\\n\"\n",
    "     \"Valid_pixels = \", format(X_valid.shape) + \"\\n\"\n",
    "     \"Valid_o_label = \", format(y_valid.shape) + \"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "VmwyoRffi350"
   },
   "source": [
    "## Feature Scaling, PCA decomposition and Model Building"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vd_sXnHBi350"
   },
   "source": [
    "Since we have a high number of dimensions, it makes sense to reduce remove any dimensions/features which cause a high amount of redundancy and focus on the dimensions which are actually required to predict the output. Before applying PCA, it is necessary we standardize both the training and validation datasets."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "DONKdV93i350"
   },
   "source": [
    "### Min-Max Scaling on Train and Validation datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "id": "QunRVyTui350"
   },
   "outputs": [],
   "source": [
    "'''Before applying PCA'''\n",
    "\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "#Scaling data\n",
    "scale = MinMaxScaler()\n",
    "\n",
    "# Scaling train data-set\n",
    "X_train_scaled = scale.fit_transform(X_train)\n",
    "\n",
    "\n",
    "# Scaling validation data-set\n",
    "X_valid_scaled = scale.fit_transform(X_valid)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "4wmtQBsqi350"
   },
   "source": [
    "### PCA Decomposition"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Principal component analysis is linear dimensionality reduction using single value decomposition of data to lower the number of dimensions. Choose the n_components parameter of PCA which gives higher accuracy score for the dataset and remove the dimensions which cause high redundancy accordingly to focus on dimensions which are actually required"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "rRq-Mwqti351"
   },
   "outputs": [],
   "source": [
    "'''function to run PCA and SVC with different kernel types'''\n",
    "from sklearn.decomposition import PCA\n",
    "# pca = decomposition.PCA()\n",
    "def n_components_accuracy(n,X_train, y_train, X_valid, y_valid, k_type):\n",
    "    pca = PCA(n_components = n)\n",
    "    print(\"n_components = \", format(n))\n",
    "    pca.fit(X_train)\n",
    "    pca_X_train = pca.transform(X_train)\n",
    "    pca_X_val = pca.transform(X_valid)\n",
    "    \n",
    "    svc_applied = SVC(kernel = k_type)\n",
    "    svc_applied.fit(pca_X_train,y_train)\n",
    "    svc_predict_label = svc_applied.predict(pca_X_val)\n",
    "    accuracy_svc = metrics.accuracy_score(y_valid,svc_predict_label)\n",
    "    print(\"Accuracy = \", format(accuracy_svc))\n",
    "    return accuracy_svc\n",
    "\n",
    "# pca = decomposition.PCA()\n",
    "# pca.n_components = 2\n",
    "# pca_data_train = pca.fit_transform(X_train)\n",
    "# pca_data_valid = pca.fit_transform(X_valid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "id": "Vg5KN5zbi351"
   },
   "outputs": [],
   "source": [
    "'''we typically want the 85-95% of the data explaining the variance. So we check for explaination of variance \n",
    "when n_components are from .85 to .95'''\n",
    "\n",
    "n_component_list = np.linspace(0.85,0.95,num=10)\n",
    "accuracy = []\n",
    "def pca_svc(n_component_list,kernel_name):\n",
    "    print(\"Running SVC with kernel type = \", format(kernel_name))\n",
    "    for n in n_component_list:\n",
    "        acc= n_components_accuracy(n,X_train_scaled, y_train, X_valid_scaled, y_valid, kernel_name)\n",
    "        accuracy.append(acc)\n",
    "    \n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "id": "UQmjvtaxi351"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running SVC with kernel type =  linear\n",
      "n_components =  0.85\n",
      "Accuracy =  0.9332142857142857\n",
      "n_components =  0.861111111111111\n",
      "Accuracy =  0.934404761904762\n",
      "n_components =  0.8722222222222222\n",
      "Accuracy =  0.9353571428571429\n",
      "n_components =  0.8833333333333333\n",
      "Accuracy =  0.9365476190476191\n",
      "n_components =  0.8944444444444444\n",
      "Accuracy =  0.9371428571428572\n",
      "n_components =  0.9055555555555556\n",
      "Accuracy =  0.9357142857142857\n",
      "n_components =  0.9166666666666666\n",
      "Accuracy =  0.9372619047619047\n",
      "n_components =  0.9277777777777777\n",
      "Accuracy =  0.9371428571428572\n",
      "n_components =  0.9388888888888889\n",
      "Accuracy =  0.9375\n",
      "n_components =  0.95\n",
      "Accuracy =  0.9348809523809524\n"
     ]
    }
   ],
   "source": [
    "pca_svc(n_component_list,\"linear\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "id": "es7jDpXQi351"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Running SVC with kernel type =  rbf\n",
      "n_components =  0.85\n",
      "Accuracy =  0.9811904761904762\n",
      "n_components =  0.861111111111111\n",
      "Accuracy =  0.9814285714285714\n",
      "n_components =  0.8722222222222222\n",
      "Accuracy =  0.9816666666666667\n",
      "n_components =  0.8833333333333333\n",
      "Accuracy =  0.9811904761904762\n",
      "n_components =  0.8944444444444444\n",
      "Accuracy =  0.9811904761904762\n",
      "n_components =  0.9055555555555556\n",
      "Accuracy =  0.9808333333333333\n",
      "n_components =  0.9166666666666666\n",
      "Accuracy =  0.9809523809523809\n",
      "n_components =  0.9277777777777777\n",
      "Accuracy =  0.9810714285714286\n",
      "n_components =  0.9388888888888889\n",
      "Accuracy =  0.9811904761904762\n",
      "n_components =  0.95\n",
      "Accuracy =  0.9809523809523809\n"
     ]
    }
   ],
   "source": [
    "del accuracy\n",
    "accuracy = []\n",
    "pca_svc(n_component_list,\"rbf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "id": "x9gEBAQ_i351"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plt.plot(n_component_list,np.array(accuracy),label='Accuracy vs % variance explained')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "D501V0H_i352"
   },
   "source": [
    "From the above we can see that the accuracy peaked around 0.87. So, we are considering n_components for PCA to be 0.87"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ei7W3msdi352"
   },
   "source": [
    "With n_components as 0.87, the RBF model gives the highest accuracy - we will proceed with this model.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "-pyqKnfii352"
   },
   "source": [
    "### Optimising hyperparameter (Using Gridsearch)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "id": "RwtdnMH2i352",
    "outputId": "624b50b2-337b-416c-db07-4f0f2247d14c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 4 folds for each of 9 candidates, totalling 36 fits\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=4, estimator=SVC(), n_jobs=-1,\n",
       "             param_grid=[{'C': [1, 10, 20], 'gamma': [0.01, 0.001, 0.0001]}],\n",
       "             return_train_score=True, scoring='accuracy', verbose=1)"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# #Creating a KFold object with 5 splits \n",
    "# folds = KFold(n_splits = 5, shuffle = True, random_state = 101)\n",
    "\n",
    "# #Setting the parameters for cross-validation\n",
    "# hyper_params = [ {'gamma': [1e-2, 1e-3, 1e-4],\n",
    "#                     'C': [1, 10, 20]}]\n",
    "\n",
    "\n",
    "# #Specifying kernel to be used for model building\n",
    "# model = SVC(kernel=\"rbf\")\n",
    "\n",
    "# #Setting up GridSearchCV\n",
    "# model_cv = GridSearchCV(estimator = model, \n",
    "#                        param_grid = hyper_params, \n",
    "#                        scoring= 'accuracy', \n",
    "#                        cv = 4,\n",
    "#                        n_jobs=-1,\n",
    "#                        verbose = 1,\n",
    "#                        return_train_score=True)      \n",
    "\n",
    "# #Fitting the model\n",
    "# model_cv.fit(X_train_scaled, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "id": "RStOaSboi352"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>mean_fit_time</th>\n",
       "      <th>std_fit_time</th>\n",
       "      <th>mean_score_time</th>\n",
       "      <th>std_score_time</th>\n",
       "      <th>param_C</th>\n",
       "      <th>param_gamma</th>\n",
       "      <th>params</th>\n",
       "      <th>split0_test_score</th>\n",
       "      <th>split1_test_score</th>\n",
       "      <th>split2_test_score</th>\n",
       "      <th>split3_test_score</th>\n",
       "      <th>mean_test_score</th>\n",
       "      <th>std_test_score</th>\n",
       "      <th>rank_test_score</th>\n",
       "      <th>split0_train_score</th>\n",
       "      <th>split1_train_score</th>\n",
       "      <th>split2_train_score</th>\n",
       "      <th>split3_train_score</th>\n",
       "      <th>mean_train_score</th>\n",
       "      <th>std_train_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>188.763210</td>\n",
       "      <td>2.122584</td>\n",
       "      <td>107.841897</td>\n",
       "      <td>2.962313</td>\n",
       "      <td>1</td>\n",
       "      <td>0.01</td>\n",
       "      <td>{'C': 1, 'gamma': 0.01}</td>\n",
       "      <td>0.969524</td>\n",
       "      <td>0.966786</td>\n",
       "      <td>0.971310</td>\n",
       "      <td>0.965357</td>\n",
       "      <td>0.968244</td>\n",
       "      <td>0.002318</td>\n",
       "      <td>3</td>\n",
       "      <td>0.982738</td>\n",
       "      <td>0.983175</td>\n",
       "      <td>0.983056</td>\n",
       "      <td>0.983016</td>\n",
       "      <td>0.982996</td>\n",
       "      <td>0.000160</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>337.558379</td>\n",
       "      <td>2.436431</td>\n",
       "      <td>144.662939</td>\n",
       "      <td>2.490811</td>\n",
       "      <td>1</td>\n",
       "      <td>0.001</td>\n",
       "      <td>{'C': 1, 'gamma': 0.001}</td>\n",
       "      <td>0.929405</td>\n",
       "      <td>0.919286</td>\n",
       "      <td>0.932143</td>\n",
       "      <td>0.926071</td>\n",
       "      <td>0.926726</td>\n",
       "      <td>0.004804</td>\n",
       "      <td>7</td>\n",
       "      <td>0.931151</td>\n",
       "      <td>0.933651</td>\n",
       "      <td>0.931389</td>\n",
       "      <td>0.932659</td>\n",
       "      <td>0.932212</td>\n",
       "      <td>0.001009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>862.029877</td>\n",
       "      <td>2.948528</td>\n",
       "      <td>282.529215</td>\n",
       "      <td>7.055029</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0001</td>\n",
       "      <td>{'C': 1, 'gamma': 0.0001}</td>\n",
       "      <td>0.888333</td>\n",
       "      <td>0.880119</td>\n",
       "      <td>0.889405</td>\n",
       "      <td>0.890357</td>\n",
       "      <td>0.887054</td>\n",
       "      <td>0.004067</td>\n",
       "      <td>9</td>\n",
       "      <td>0.888373</td>\n",
       "      <td>0.891389</td>\n",
       "      <td>0.887579</td>\n",
       "      <td>0.888571</td>\n",
       "      <td>0.888978</td>\n",
       "      <td>0.001440</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>164.410081</td>\n",
       "      <td>2.849948</td>\n",
       "      <td>98.758446</td>\n",
       "      <td>1.915554</td>\n",
       "      <td>10</td>\n",
       "      <td>0.01</td>\n",
       "      <td>{'C': 10, 'gamma': 0.01}</td>\n",
       "      <td>0.977381</td>\n",
       "      <td>0.975476</td>\n",
       "      <td>0.978929</td>\n",
       "      <td>0.974762</td>\n",
       "      <td>0.976637</td>\n",
       "      <td>0.001633</td>\n",
       "      <td>1</td>\n",
       "      <td>0.999643</td>\n",
       "      <td>0.999683</td>\n",
       "      <td>0.999722</td>\n",
       "      <td>0.999762</td>\n",
       "      <td>0.999702</td>\n",
       "      <td>0.000044</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>168.040193</td>\n",
       "      <td>2.287961</td>\n",
       "      <td>97.158844</td>\n",
       "      <td>7.388441</td>\n",
       "      <td>10</td>\n",
       "      <td>0.001</td>\n",
       "      <td>{'C': 10, 'gamma': 0.001}</td>\n",
       "      <td>0.950238</td>\n",
       "      <td>0.943571</td>\n",
       "      <td>0.951071</td>\n",
       "      <td>0.946667</td>\n",
       "      <td>0.947887</td>\n",
       "      <td>0.002991</td>\n",
       "      <td>5</td>\n",
       "      <td>0.960040</td>\n",
       "      <td>0.960913</td>\n",
       "      <td>0.960714</td>\n",
       "      <td>0.961587</td>\n",
       "      <td>0.960813</td>\n",
       "      <td>0.000552</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   mean_fit_time  std_fit_time  mean_score_time  std_score_time param_C  \\\n",
       "0     188.763210      2.122584       107.841897        2.962313       1   \n",
       "1     337.558379      2.436431       144.662939        2.490811       1   \n",
       "2     862.029877      2.948528       282.529215        7.055029       1   \n",
       "3     164.410081      2.849948        98.758446        1.915554      10   \n",
       "4     168.040193      2.287961        97.158844        7.388441      10   \n",
       "\n",
       "  param_gamma                     params  split0_test_score  \\\n",
       "0        0.01    {'C': 1, 'gamma': 0.01}           0.969524   \n",
       "1       0.001   {'C': 1, 'gamma': 0.001}           0.929405   \n",
       "2      0.0001  {'C': 1, 'gamma': 0.0001}           0.888333   \n",
       "3        0.01   {'C': 10, 'gamma': 0.01}           0.977381   \n",
       "4       0.001  {'C': 10, 'gamma': 0.001}           0.950238   \n",
       "\n",
       "   split1_test_score  split2_test_score  split3_test_score  mean_test_score  \\\n",
       "0           0.966786           0.971310           0.965357         0.968244   \n",
       "1           0.919286           0.932143           0.926071         0.926726   \n",
       "2           0.880119           0.889405           0.890357         0.887054   \n",
       "3           0.975476           0.978929           0.974762         0.976637   \n",
       "4           0.943571           0.951071           0.946667         0.947887   \n",
       "\n",
       "   std_test_score  rank_test_score  split0_train_score  split1_train_score  \\\n",
       "0        0.002318                3            0.982738            0.983175   \n",
       "1        0.004804                7            0.931151            0.933651   \n",
       "2        0.004067                9            0.888373            0.891389   \n",
       "3        0.001633                1            0.999643            0.999683   \n",
       "4        0.002991                5            0.960040            0.960913   \n",
       "\n",
       "   split2_train_score  split3_train_score  mean_train_score  std_train_score  \n",
       "0            0.983056            0.983016          0.982996         0.000160  \n",
       "1            0.931389            0.932659          0.932212         0.001009  \n",
       "2            0.887579            0.888571          0.888978         0.001440  \n",
       "3            0.999722            0.999762          0.999702         0.000044  \n",
       "4            0.960714            0.961587          0.960813         0.000552  "
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# cv_results = pd.DataFrame(model_cv.cv_results_)\n",
    "# cv_results.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 187
    },
    "id": "MDNR2HKoi352",
    "outputId": "a98ba9ec-703a-4e8b-ba7a-94507c083fa8"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>mean_fit_time</th>\n",
       "      <th>std_fit_time</th>\n",
       "      <th>mean_score_time</th>\n",
       "      <th>std_score_time</th>\n",
       "      <th>param_C</th>\n",
       "      <th>param_gamma</th>\n",
       "      <th>params</th>\n",
       "      <th>split0_test_score</th>\n",
       "      <th>split1_test_score</th>\n",
       "      <th>split2_test_score</th>\n",
       "      <th>split3_test_score</th>\n",
       "      <th>mean_test_score</th>\n",
       "      <th>std_test_score</th>\n",
       "      <th>rank_test_score</th>\n",
       "      <th>split0_train_score</th>\n",
       "      <th>split1_train_score</th>\n",
       "      <th>split2_train_score</th>\n",
       "      <th>split3_train_score</th>\n",
       "      <th>mean_train_score</th>\n",
       "      <th>std_train_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>164.410081</td>\n",
       "      <td>2.849948</td>\n",
       "      <td>98.758446</td>\n",
       "      <td>1.915554</td>\n",
       "      <td>10</td>\n",
       "      <td>0.01</td>\n",
       "      <td>{'C': 10, 'gamma': 0.01}</td>\n",
       "      <td>0.977381</td>\n",
       "      <td>0.975476</td>\n",
       "      <td>0.978929</td>\n",
       "      <td>0.974762</td>\n",
       "      <td>0.976637</td>\n",
       "      <td>0.001633</td>\n",
       "      <td>1</td>\n",
       "      <td>0.999643</td>\n",
       "      <td>0.999683</td>\n",
       "      <td>0.999722</td>\n",
       "      <td>0.999762</td>\n",
       "      <td>0.999702</td>\n",
       "      <td>0.000044</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>163.143296</td>\n",
       "      <td>3.621349</td>\n",
       "      <td>91.512228</td>\n",
       "      <td>2.638200</td>\n",
       "      <td>20</td>\n",
       "      <td>0.01</td>\n",
       "      <td>{'C': 20, 'gamma': 0.01}</td>\n",
       "      <td>0.977857</td>\n",
       "      <td>0.975238</td>\n",
       "      <td>0.978333</td>\n",
       "      <td>0.975000</td>\n",
       "      <td>0.976607</td>\n",
       "      <td>0.001500</td>\n",
       "      <td>2</td>\n",
       "      <td>0.999921</td>\n",
       "      <td>0.999960</td>\n",
       "      <td>0.999960</td>\n",
       "      <td>0.999960</td>\n",
       "      <td>0.999950</td>\n",
       "      <td>0.000017</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>188.763210</td>\n",
       "      <td>2.122584</td>\n",
       "      <td>107.841897</td>\n",
       "      <td>2.962313</td>\n",
       "      <td>1</td>\n",
       "      <td>0.01</td>\n",
       "      <td>{'C': 1, 'gamma': 0.01}</td>\n",
       "      <td>0.969524</td>\n",
       "      <td>0.966786</td>\n",
       "      <td>0.971310</td>\n",
       "      <td>0.965357</td>\n",
       "      <td>0.968244</td>\n",
       "      <td>0.002318</td>\n",
       "      <td>3</td>\n",
       "      <td>0.982738</td>\n",
       "      <td>0.983175</td>\n",
       "      <td>0.983056</td>\n",
       "      <td>0.983016</td>\n",
       "      <td>0.982996</td>\n",
       "      <td>0.000160</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>152.804434</td>\n",
       "      <td>3.021952</td>\n",
       "      <td>86.967303</td>\n",
       "      <td>4.537312</td>\n",
       "      <td>20</td>\n",
       "      <td>0.001</td>\n",
       "      <td>{'C': 20, 'gamma': 0.001}</td>\n",
       "      <td>0.953095</td>\n",
       "      <td>0.949762</td>\n",
       "      <td>0.955952</td>\n",
       "      <td>0.949643</td>\n",
       "      <td>0.952113</td>\n",
       "      <td>0.002614</td>\n",
       "      <td>4</td>\n",
       "      <td>0.970714</td>\n",
       "      <td>0.971190</td>\n",
       "      <td>0.970635</td>\n",
       "      <td>0.971389</td>\n",
       "      <td>0.970982</td>\n",
       "      <td>0.000317</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>168.040193</td>\n",
       "      <td>2.287961</td>\n",
       "      <td>97.158844</td>\n",
       "      <td>7.388441</td>\n",
       "      <td>10</td>\n",
       "      <td>0.001</td>\n",
       "      <td>{'C': 10, 'gamma': 0.001}</td>\n",
       "      <td>0.950238</td>\n",
       "      <td>0.943571</td>\n",
       "      <td>0.951071</td>\n",
       "      <td>0.946667</td>\n",
       "      <td>0.947887</td>\n",
       "      <td>0.002991</td>\n",
       "      <td>5</td>\n",
       "      <td>0.960040</td>\n",
       "      <td>0.960913</td>\n",
       "      <td>0.960714</td>\n",
       "      <td>0.961587</td>\n",
       "      <td>0.960813</td>\n",
       "      <td>0.000552</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>279.309089</td>\n",
       "      <td>1.324996</td>\n",
       "      <td>127.028254</td>\n",
       "      <td>12.954888</td>\n",
       "      <td>20</td>\n",
       "      <td>0.0001</td>\n",
       "      <td>{'C': 20, 'gamma': 0.0001}</td>\n",
       "      <td>0.933929</td>\n",
       "      <td>0.921190</td>\n",
       "      <td>0.935119</td>\n",
       "      <td>0.929524</td>\n",
       "      <td>0.929940</td>\n",
       "      <td>0.005465</td>\n",
       "      <td>6</td>\n",
       "      <td>0.935913</td>\n",
       "      <td>0.938056</td>\n",
       "      <td>0.935159</td>\n",
       "      <td>0.935595</td>\n",
       "      <td>0.936181</td>\n",
       "      <td>0.001115</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>337.558379</td>\n",
       "      <td>2.436431</td>\n",
       "      <td>144.662939</td>\n",
       "      <td>2.490811</td>\n",
       "      <td>1</td>\n",
       "      <td>0.001</td>\n",
       "      <td>{'C': 1, 'gamma': 0.001}</td>\n",
       "      <td>0.929405</td>\n",
       "      <td>0.919286</td>\n",
       "      <td>0.932143</td>\n",
       "      <td>0.926071</td>\n",
       "      <td>0.926726</td>\n",
       "      <td>0.004804</td>\n",
       "      <td>7</td>\n",
       "      <td>0.931151</td>\n",
       "      <td>0.933651</td>\n",
       "      <td>0.931389</td>\n",
       "      <td>0.932659</td>\n",
       "      <td>0.932212</td>\n",
       "      <td>0.001009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>331.623591</td>\n",
       "      <td>11.389338</td>\n",
       "      <td>158.555682</td>\n",
       "      <td>8.964112</td>\n",
       "      <td>10</td>\n",
       "      <td>0.0001</td>\n",
       "      <td>{'C': 10, 'gamma': 0.0001}</td>\n",
       "      <td>0.926071</td>\n",
       "      <td>0.915357</td>\n",
       "      <td>0.928690</td>\n",
       "      <td>0.923571</td>\n",
       "      <td>0.923423</td>\n",
       "      <td>0.004996</td>\n",
       "      <td>8</td>\n",
       "      <td>0.927738</td>\n",
       "      <td>0.929603</td>\n",
       "      <td>0.927341</td>\n",
       "      <td>0.928452</td>\n",
       "      <td>0.928284</td>\n",
       "      <td>0.000860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>862.029877</td>\n",
       "      <td>2.948528</td>\n",
       "      <td>282.529215</td>\n",
       "      <td>7.055029</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0001</td>\n",
       "      <td>{'C': 1, 'gamma': 0.0001}</td>\n",
       "      <td>0.888333</td>\n",
       "      <td>0.880119</td>\n",
       "      <td>0.889405</td>\n",
       "      <td>0.890357</td>\n",
       "      <td>0.887054</td>\n",
       "      <td>0.004067</td>\n",
       "      <td>9</td>\n",
       "      <td>0.888373</td>\n",
       "      <td>0.891389</td>\n",
       "      <td>0.887579</td>\n",
       "      <td>0.888571</td>\n",
       "      <td>0.888978</td>\n",
       "      <td>0.001440</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   mean_fit_time  std_fit_time  mean_score_time  std_score_time param_C  \\\n",
       "3     164.410081      2.849948        98.758446        1.915554      10   \n",
       "6     163.143296      3.621349        91.512228        2.638200      20   \n",
       "0     188.763210      2.122584       107.841897        2.962313       1   \n",
       "7     152.804434      3.021952        86.967303        4.537312      20   \n",
       "4     168.040193      2.287961        97.158844        7.388441      10   \n",
       "8     279.309089      1.324996       127.028254       12.954888      20   \n",
       "1     337.558379      2.436431       144.662939        2.490811       1   \n",
       "5     331.623591     11.389338       158.555682        8.964112      10   \n",
       "2     862.029877      2.948528       282.529215        7.055029       1   \n",
       "\n",
       "  param_gamma                      params  split0_test_score  \\\n",
       "3        0.01    {'C': 10, 'gamma': 0.01}           0.977381   \n",
       "6        0.01    {'C': 20, 'gamma': 0.01}           0.977857   \n",
       "0        0.01     {'C': 1, 'gamma': 0.01}           0.969524   \n",
       "7       0.001   {'C': 20, 'gamma': 0.001}           0.953095   \n",
       "4       0.001   {'C': 10, 'gamma': 0.001}           0.950238   \n",
       "8      0.0001  {'C': 20, 'gamma': 0.0001}           0.933929   \n",
       "1       0.001    {'C': 1, 'gamma': 0.001}           0.929405   \n",
       "5      0.0001  {'C': 10, 'gamma': 0.0001}           0.926071   \n",
       "2      0.0001   {'C': 1, 'gamma': 0.0001}           0.888333   \n",
       "\n",
       "   split1_test_score  split2_test_score  split3_test_score  mean_test_score  \\\n",
       "3           0.975476           0.978929           0.974762         0.976637   \n",
       "6           0.975238           0.978333           0.975000         0.976607   \n",
       "0           0.966786           0.971310           0.965357         0.968244   \n",
       "7           0.949762           0.955952           0.949643         0.952113   \n",
       "4           0.943571           0.951071           0.946667         0.947887   \n",
       "8           0.921190           0.935119           0.929524         0.929940   \n",
       "1           0.919286           0.932143           0.926071         0.926726   \n",
       "5           0.915357           0.928690           0.923571         0.923423   \n",
       "2           0.880119           0.889405           0.890357         0.887054   \n",
       "\n",
       "   std_test_score  rank_test_score  split0_train_score  split1_train_score  \\\n",
       "3        0.001633                1            0.999643            0.999683   \n",
       "6        0.001500                2            0.999921            0.999960   \n",
       "0        0.002318                3            0.982738            0.983175   \n",
       "7        0.002614                4            0.970714            0.971190   \n",
       "4        0.002991                5            0.960040            0.960913   \n",
       "8        0.005465                6            0.935913            0.938056   \n",
       "1        0.004804                7            0.931151            0.933651   \n",
       "5        0.004996                8            0.927738            0.929603   \n",
       "2        0.004067                9            0.888373            0.891389   \n",
       "\n",
       "   split2_train_score  split3_train_score  mean_train_score  std_train_score  \n",
       "3            0.999722            0.999762          0.999702         0.000044  \n",
       "6            0.999960            0.999960          0.999950         0.000017  \n",
       "0            0.983056            0.983016          0.982996         0.000160  \n",
       "7            0.970635            0.971389          0.970982         0.000317  \n",
       "4            0.960714            0.961587          0.960813         0.000552  \n",
       "8            0.935159            0.935595          0.936181         0.001115  \n",
       "1            0.931389            0.932659          0.932212         0.001009  \n",
       "5            0.927341            0.928452          0.928284         0.000860  \n",
       "2            0.887579            0.888571          0.888978         0.001440  "
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# cv_results.sort_values('mean_test_score',ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "R0N-dyV5i353"
   },
   "source": [
    "Based on the above, the optimal hyper parameters for SVC are C = 10 and gamma = 0.01"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "id": "7KPObrYui353"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The best test score is 0.9766369047619048 corresponding to hyperparameters {'C': 10, 'gamma': 0.01}\n"
     ]
    }
   ],
   "source": [
    "# # Optimal accuracy score and hyperparameters\n",
    "# best_score = model_cv.best_score_\n",
    "# best_hyperparams = model_cv.best_params_\n",
    "\n",
    "# print(\"The best test score is {0} corresponding to hyperparameters {1}\".format(best_score, best_hyperparams))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "***Above code is commented as the Gridsearch will take a longer time to find the optimal hyperparamaters. Please feel free to uncommented the code if necessary.***"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "9H73zEgZi353"
   },
   "source": [
    "### Building the model with optimal hyperparameters\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "id": "tJPwim-3i353"
   },
   "outputs": [],
   "source": [
    "# Model with optimal hyperparameters\n",
    "pca = PCA(n_components = 0.87)\n",
    "pca.fit(X_train)\n",
    "pca_X_train = pca.transform(X_train_scaled)\n",
    "pca_X_val = pca.transform(X_valid_scaled)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "id": "JOi3MKEXi353",
    "outputId": "762dd725-e35d-4bd5-d400-7f6246e279bb"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SVC(C=10, gamma=0.01)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_final = SVC(C=10, gamma=0.01, kernel=\"rbf\")\n",
    "model_final.fit(pca_X_train, y_train) # Fitting the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "id": "KsVlXFi_i353"
   },
   "outputs": [],
   "source": [
    "#model_final.fit(pca_X_val, y_train)\n",
    "# Prediction on train dataset for Non-linear model\n",
    "y_train_pred_label = model_final.predict(pca_X_train)\n",
    "\n",
    "# Prediction on validation dataset for Non-linear model\n",
    "y_valid_pred_label = model_final.predict(pca_X_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "id": "CkHXCGW1i353",
    "outputId": "783ba1f8-9688-4a82-8d77-e4c58be52002"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9972321428571429"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics.accuracy_score(y_train, y_train_pred_label)#accuracy on train data "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "id": "GnH2zsJpi353",
    "outputId": "3e8b70f1-0251-41d0-baed-e1ddadb71cd9"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9810714285714286"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics.accuracy_score(y_valid, y_valid_pred_label)#accuracy on validation data "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "id": "IgsXD32Fi353",
    "outputId": "f13c62a6-53a9-41d1-d20c-2c27710d17f2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.98      0.99      0.99       826\n",
      "           1       0.99      0.99      0.99       937\n",
      "           2       0.98      0.99      0.99       835\n",
      "           3       0.98      0.97      0.97       870\n",
      "           4       0.98      0.99      0.98       814\n",
      "           5       0.97      0.97      0.97       759\n",
      "           6       0.99      0.99      0.99       828\n",
      "           7       0.98      0.98      0.98       880\n",
      "           8       0.98      0.98      0.98       813\n",
      "           9       0.98      0.96      0.97       838\n",
      "\n",
      "    accuracy                           0.98      8400\n",
      "   macro avg       0.98      0.98      0.98      8400\n",
      "weighted avg       0.98      0.98      0.98      8400\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(metrics.classification_report(y_valid, y_valid_pred_label))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "C5KIuM0pi354"
   },
   "source": [
    "## Evaluating on unknown test data (from Kaggle)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "M2My7hlpi354"
   },
   "source": [
    "Considering the test.csv from kaggle website - https://www.kaggle.com/competitions/digit-recognizer/data to predict the label column. Once we have the predicted labels, the output is uploaded to Kaggle for an accuracy score. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "id": "kLcP4OJZi354",
    "outputId": "17bee06f-66fe-4677-c8c0-2528437a8c20"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>pixel0</th>\n",
       "      <th>pixel1</th>\n",
       "      <th>pixel2</th>\n",
       "      <th>pixel3</th>\n",
       "      <th>pixel4</th>\n",
       "      <th>pixel5</th>\n",
       "      <th>pixel6</th>\n",
       "      <th>pixel7</th>\n",
       "      <th>pixel8</th>\n",
       "      <th>pixel9</th>\n",
       "      <th>...</th>\n",
       "      <th>pixel774</th>\n",
       "      <th>pixel775</th>\n",
       "      <th>pixel776</th>\n",
       "      <th>pixel777</th>\n",
       "      <th>pixel778</th>\n",
       "      <th>pixel779</th>\n",
       "      <th>pixel780</th>\n",
       "      <th>pixel781</th>\n",
       "      <th>pixel782</th>\n",
       "      <th>pixel783</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 784 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  pixel8  \\\n",
       "0       0       0       0       0       0       0       0       0       0   \n",
       "1       0       0       0       0       0       0       0       0       0   \n",
       "2       0       0       0       0       0       0       0       0       0   \n",
       "3       0       0       0       0       0       0       0       0       0   \n",
       "4       0       0       0       0       0       0       0       0       0   \n",
       "\n",
       "   pixel9  ...  pixel774  pixel775  pixel776  pixel777  pixel778  pixel779  \\\n",
       "0       0  ...         0         0         0         0         0         0   \n",
       "1       0  ...         0         0         0         0         0         0   \n",
       "2       0  ...         0         0         0         0         0         0   \n",
       "3       0  ...         0         0         0         0         0         0   \n",
       "4       0  ...         0         0         0         0         0         0   \n",
       "\n",
       "   pixel780  pixel781  pixel782  pixel783  \n",
       "0         0         0         0         0  \n",
       "1         0         0         0         0  \n",
       "2         0         0         0         0  \n",
       "3         0         0         0         0  \n",
       "4         0         0         0         0  \n",
       "\n",
       "[5 rows x 784 columns]"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_data = pd.read_csv('test.csv')# downloaded from https://www.kaggle.com/c/digit-recognizer/data\n",
    "test_data.head()# this data set has no 'label' column"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "id": "XdnlqvDHi354",
    "outputId": "d2dd9ea1-2f36-4c5e-dc07-e084e730aad1"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8400, 63)"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca_X_val.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "id": "ymsea5pji354",
    "outputId": "24671f96-95b9-48ed-cb40-eae057073a3d"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(33600, 63)"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca_X_train.shape\n",
    "# pca_X_train = pd.DataFrame(pca_X_train)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "id": "D4QAtD5Wi354",
    "outputId": "3763277c-fe80-4ffe-c24c-730ef22fe47e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(33600, 63)"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca_X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "id": "AxTbRvgsi354"
   },
   "outputs": [],
   "source": [
    "test_data = test_data[X_train.columns]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "id": "uqzegMZAi355",
    "outputId": "95853f97-dd21-4867-c0e0-eee3b01aa8cb"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(33600, 784)"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_scaled.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "id": "5riZgYv0i355",
    "outputId": "07492590-8bbf-4663-a35e-a9571db75da3"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(28000, 784)"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "id": "1GV4AU4Xi355",
    "outputId": "d15ba844-5e6c-4eb0-9d20-8226b21dddbf"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(33600, 63)"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca_X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "id": "MOvrDC7Li355"
   },
   "outputs": [],
   "source": [
    "\n",
    "# mnist_test_arr = np.array(test_data)\n",
    "# mnist_test_arr = mnist_test_arr.reshape(mnist_test_arr.shape[0], 28, 28, 1)\n",
    "\n",
    "test_data_scaled = scale.transform(test_data)#feature scaling the data using minmax scaler.\n",
    "# test_data=(test_data-X.min())/(X.max()-X.min())\n",
    "# X_t=df_test[X.columns]\n",
    "# Doing the PCA transform\n",
    "# X_trans=pca.transform(X_t)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "id": "VuYWYGZmi355",
    "outputId": "e7abd269-3df9-498f-cf19-48746e298eeb"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(28000, 784)"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_data_scaled.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "id": "W4T4GXeFi355"
   },
   "outputs": [],
   "source": [
    "# pca = PCA(n_components = 0.86)\n",
    "# pca.fit(test_data)\n",
    "pca_X_test = pca.transform(test_data_scaled)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "id": "ozAa_X5Xi355",
    "outputId": "213a0ae7-42a1-4ef6-a174-daba0090d2bf"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(28000, 63)"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca_X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "id": "dM_8BWNOi355"
   },
   "outputs": [],
   "source": [
    "y_pred_final = model_final.predict(pca_X_test) #predicting on test data.."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "id": "Ugeuky1Ci355",
    "outputId": "6f01d453-fbe1-4b32-8e52-6015103e01ab"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([2, 0, 9, ..., 3, 9, 2], dtype=int64)"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred_final"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "id": "Nf8eUXbji356",
    "outputId": "4b94bc57-0063-4f40-e14f-2c4bd0cba012"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ImageId</th>\n",
       "      <th>Label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   ImageId  Label\n",
       "0        1      2\n",
       "1        2      0\n",
       "2        3      9\n",
       "3        4      9\n",
       "4        5      3"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "id = np.arange(1,28001,1)#creating a data frame with imageid and predicted label to submit it on kaggle.\n",
    "kaggle_df = pd.concat([pd.Series(id),pd.Series(y_pred_final)],axis=1)\n",
    "kaggle_df.columns = ['ImageId','Label']\n",
    "kaggle_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "id": "8VMUHWl3i356"
   },
   "outputs": [],
   "source": [
    "kaggle_df.to_csv('mnist_svm.csv',index=False)#exporting is as csv file, submitting it on kaggle and to recieve a score. "
   ]
  },
  {
   "attachments": {
    "image.png": {
     "image/png": "iVBORw0KGgoAAAANSUhEUgAABNYAAAECCAYAAAAy4lohAAAgAElEQVR4nOzde1iUdf7/8WdijAqOimJSIBlYLVgu4Ka4+k1z81DYdFi0TaIttcysPHQwysytyA6mvzxtntplYTdlS1FLwnXBcsHchFqB3QIqhJKcRJkQnUn098cMOIOgOKKj8npcl1fNzH3fn/d9zy1c98vP4ZJjx44dQ0RERERERC4IXxZ9zYEqCyG9etKtq5+ny5Ez8OO+Skq+2U3nTkau6X3VWWlD98v571zcB3L2XKJgTURERERE5MJi/rGSb0rLMHh7c8XllykwucD8uK+S777/AavNRq/gIPy7nd3vT/fL+elc3wdydihYExERERERuUDt+cHM9xU/cPiwla5+Xehs7IivTwfatTPQtm1bT5cnwJEjRzh82Er1wRoOWH5iX+V+2rUzcHmPywi4zP+c1qL7xXPOp/tAWpaCNRERERERkQtcTc0hKvcfoMpSzcGaGg5bbdQeOYIe9jzrEsCrbVvaGbzx6dCBTkZf/Lp0pkOH9h6tS/fLuXW+3gfSMhSsiYiIiIiIiIiIuKGNpwsQERERERERERG5EClYExERERERERERcYOCNRERERERERERETcoWBMREREREREREXGDgjURERERERERERE3KFgTERERERERERFxg4I1ERERERERERERNyhYExERERERERERcYOCNRERERERERERETcoWBMREREREREREXGDgjURERERERERERE3KFgTERERERERERFxg4I1ERERERERERERNyhYExERERERERERcYOCNRERERERERERETcoWBMREREREREREXGDgjURERERERERERE3KFgTERERERERERFxg4I1ERERERERERERNyhYExERERERERERcYOCNRERERERERERETcoWBMREREREREREXGDgjURERERERERERE3tPV0Ac119OhRDh48RM3hw1itNo4cOUJt7VFPlyUiIiIiIiIi0ixeXm1o27YtBoM3Hdq1w8enPW3aqM/TheySY8eOHfN0ESdjtf3MgSoLFks1vj4d6NC+HYZ23lza9lK8vHTziYiIiIiIiMiFobb2KD8f+RnrYRs1hw5TfbAGo9GXzp2MGLwv9XR54obzOlj7cd9+qizVdOlspJOxo4I0EREREREREblo1NYepcryE/sPWOhk9KVb1y6eLklO03kZrFltP/PD3h8xeHvTrWsXBWoiIiIiIiIictGqrT3Kj/v2Y7XZuKx7N/Veu4Ccd8HaocNW9lTspWuXznTq1NHT5YiIiIiIiIiInBNVVT+xb/8BAnp0p307g6fLkWY4r7qCWW0/s6diL/7d/BSqiYiIiIiIiEir0qlTR/y7+bGnYi9W28+eLkea4bwK1n7Y+yNdu3Smo6+Pp0sRERERERERETnnOvr60LVLZ37Y+6OnS5FmOG+CtR/37cfg7a2eaiIiIiIiIiLSqnXq1BGDtzc/7tvv6VLkFM6LYM1q+5kqS7VWvxARERERERERAbp17UKVpVpDQs9z50WwdqDKQpfORq3+KSIiIiIiIiICeHm1oUtnIweqLJ4uRU7C40nW0aNHsViq6WTUEFARERERERERkTqdjB2xWKo5evSop0uRJng8WDt48BC+Ph3UW01ERERERERExImXVxt8fTpw8OAhT5ciTfB4mlVz+DAd2rfzdBkiIiIiIiIiIuedDu3bUXP4sKfLkCZ4PFizWm0Y2nl7ugwRERERERERkfOOoZ03VqvN02VIEzwerB05coRL217q6TJERERERERERM47l7a9lCNHjni6DGmCx4O12tqjml9NRERERERERKQRXl5tqK3V4gXnKyVaIiIiIiIiIiIiblCwJiIiIiIiIiIi4gYFayIiIiIiIiIiIm5QsCYiIiIiIiIiIuIGBWsiIiIiIiIiIiJuULAmIiIiIiIiIiLiBgVrIiIiIiIiIiIiblCwJiIiIiIiIiIi4gYFayIiIiIiIiIiIm5QsCYiIiIiIiIiIuIGBWsiIiIiIiIiIiJuULAmIiIiIiIiIiLiBgVrIiIiIiIiIiIibmjr6QLOulorFosVg9GIwevk75m/LaRoj4GgyDCCOjQ8Bhi7GI6/V2PBghGjYzvrfgvWus+8DBiNTts2cNJtrRYsNa7bu9QJUFNOQW4Z1oBQwq/0d/nM5dgNjm/db4ETrsPx87JaLFhrG1ZrcD1vZ7VWygpzKavqRNCvwghyPo3TbKvhOVr3W7Aajl/fpt4/oR3AUpJH/h7oHh5BaJcGNVvLKfh3GVWdgogKC2yyvnonu9YWRy0G52NYT/y+REREREREROSidPH3WPOysGlGFDHLCuvfKk6+l74T1lNW94Y5i0RTFDeMeYbEuRO4uU8UD68tP36MvPn0jZpPrtNhc9+Kou9beXUbMC8qisFj7uWe39/Lb37dh75TNlB2Qkh16m3NG2bQ9/9Gcs/v763/k/Sfun2tFKc8Qt8+I7l/7lyeHjOQvqb55NY0fux7fn8v98zditnps74zMrA0el5mMuc69rl1oFMNyS7nXa8mj8SRfYiZPtd+zfqZWJRXF+nZ25qX57R9g7bWTY3ihlud2jMtIbc+EXTUOnIJBc7XsDSZe6JOvO717dSWs3piH2649wUSX3+Gu6KieHiDuX5362dzufkXJqa+PpfEh0a6tnnCd3yqa20/h74xTjWa1zM1agbrjjcpIiIiIiIiIhexi7/HGv6MfWIaK2Lms/qO5Yw1ZLDoDQtTkmIJ9QKwkP7yFJIiFvFF2hCMXvYAJmbMdJIi1xAf3PyWxr6SRkI/YP8GHo56k7TJo5kS5sa2v5rGO6ti8W+4U+FKHp5lZmLaTqZcZ3AESUOZ+ucRfPxw2InHboR1/XSeH57NgluMJ1ynkYlpjATMf5/IDR8Ob7wGB/OHS1jePZGPU2IJAorffYSpH+QQHzGEhkduysDpK1j1W3+ozSGxdzzpuyYT6Vx3+RLW5IxnziB7l7CC9SsbD/nq5CUzc0scKUUzGegF1m1zuWvVRxSPjiMUM+uWrKR7YiYpdwdCbQmrp0wnPSeOyCGNVOy41lPS85l4NfZrPWkk97w+gC9mR1Pfr61kPlOXDWHzw0180SIiIiIiIiJy0br4e6wBhI3n1QeLmb14A5lvv0L67TOZEuGIRqy55Kz3Z8rv7KEagKFfHBOi88jKdbPrkaEjnTBjtZ5600a3tVrZu9+CZb8Fy/7jHxT/+yOKb5tA/HWO2r0CGbt0JxvHhbgc0lpVt68FS4MaJj45mYKnZ5BW4d6p1TH2CMSQs5LFKTkUm62E3r2Yjc81P1QDsB40Y9lvwZyzlcwusUS6ngam24aT9K6jh501hzXLwhh5y0kO6B9IJMnMe2sDBaVWDINmsnFVHKH2igkKNpC9cglJ20owHwlh7NI0EhoL1Th+rcde7XjDK5CxD8Rh/ft2Cpy2GzptGuFLpjMvrzlftoiIiIiIiIhcTFpHsIaByIeeYeS66TyQEsqrjww53uNov5kyQvH3c90eA9DoUM6mrX7GRIzJxM2DJ7Ju2GzGXO/mtjlziImKom9UlMvwRMueQujh7xpeGYwnzOeWNLFu3yimbmgQDv5qMkufqOLpWamU/Xx65+fS7KDZfPLeeIzb5nLPTX24NiaB9PJT7+ds59vPcM/v7+X+OanwmwhCG0xv1v32WOK3/oV1pWDJSGb1b+MYE3iSAwbH8dfMRYwsT2XqnX3oFRXP8l11gZeBgc9l8t5DRnLm3svgX/QhJiGjieG6TVzrS4GDDQK0gFheXRjN8qfmk2tBRERERERERFqRVhKsAV2GEz8OGPsAph5O7/uHEuqTReH/nAKTWjPmYjB2cE16DjuFMIePnNjEwN/NJOGhIRj3hzHjyViCTjKB/Um3HZLIjq+L+ObrIr75eiaRjreDQodATgHFTptavsohe5fZZcGCiWvq9i2yD7VsIDTuTebUzmH2qpKmCzyVGguGa2JJWJrGjs93smpADg/P2YBLjPdzE//vMHD6CjampbFxcyZTDiYwNbVBPb7RjHnQzIr1qaz701bGDouk08lqslqwdolm4rwkNu8s4uPpBhLvWVIfTFotBsLvmMnSjdn87/PlDNz2CLPXN94rsdFrvacMAo00XMrBMGQmq278iKdfz2DvyeoTERERERERkYtK6wnW6rRt5/raKwLT5BCSXp5LZrkVrGYKkuezaH8cY4Y5+itdHclYn2RWpJZgrQXrV6msWG1gbITr2MWg66MZeMtknn3QwrwlTosENOKk27oMBT2+eqb/sFhMX88l8c+FWGrBWprFvEcmsqIYl7DHZShow7GgYB/W+IfZsD2rmRftRAV/NtF3SjLFjsNbAXwMjjpCiIw1kLQqleIaoKaE1auSMcRGOoZlurKaSyjbA5aaE2sNv208/vMTmG2ezJjopldaBbBseYG+MfPJ3A/UWu01damrqZDlpigeftf+HVJrb8toaPyYjV7rtzKIfOg2wk/Y2sDA6W8y9Nssl2GiIiIiIiIiInJxa33BWiPCH/wLKbeVMPX/+tDrFwO5Z30gr26czcC6zMU4nDl/nYb19ZFc27s3146ci/XRv5Dwm8bm53IMO93yCotyTjXvVhPbugwFdVr1sstwXv3rNHh7DH179+bamATKYtew9A7XXmnOQ0H7Tl1Po32yAmOZM3vICb2vmiv8vsUkeC0h5he96dU7ioe3R7P0ieGOoZNGRs5ewwzrXG7u05tefUaSaJ3GX58d7jK0MvOpgfS6qjfX9r+XJL+ZLL2vkQUAgm8j/hbsgdZJegACGEdMZ+mNuUyN6k2v3n24+U0rCYvHO4KwMCYungmLTVzbuze9oqaQPWgxM0Y0MStcl+G8+t5suqe6Xut37m5iLGqHCGa8Mr7R4FBERERERERELk6XHDt27JgnCygqKaV3yGksvelpNVbo4G4cdRGqtWI9YqCJjl+OnmMGDKcIxVqatQYMHZr60Iq17bmvSURERERERMQdF1x20oq09XQBFxyFaq68ThFQeRnc7hV3JpoM1QAMnqlJRERERERERC4uGgoqIiIiIiIiIiLiBgVrIiIiIiIiIiIiblCwJiIiIiIiIiIi4gYFayIiIiIiIiIiIm5QsCYiIiIiIiIiIuIGBWsiIiIiIiIiIiJuULAmIiIiIiIiIiLiBgVrIiIiIiIiIiIiblCwJiIiIiIiIiIi4gYFayIiIiIiIiIiIm5QsCYiIiIiIiIiIuIGBWsiIiIiIiIiIiJuULAmIiIiIiIiIiLiBgVrIiIiIiIiIiIiblCwJiIiIiIiIiIi4gYFayIiIiIiIiIiIm5QsCYiIiIiIiIiIuIGBWsiIiIiIiIiIiJuULAmIiIiIiIiIiLiBgVrIiIiIiIiIiIiblCwJiIiIiIiIiIi4gYFayIiIiIiIiIiIm5o6+kCLhg/2zB/l092YS755u3kH4K91iKKjwJt/Bho6Art+3CTf2/Cw6KJusIPw6WeLlpERERERERERM4WBWunYPl2O+u2JbHMXEBZUxsdrST7UCUcKiK7EvgSwJeh/vHcN+hWhl7pe87q9RSrxQI+Rgxenq5EREREREREROTcuOTYsWPHPFlAUUkpvUOCPVlCoyzFH7Dwn0tYdrD6jI8V5HMHT940AVPoxRmwmddOZPCMLAwPJLHjuWgMni5IRERERETkPGUpyWLdn1JZ859yIJCoMbHcO3oIoUZPV3YGai0UfLiS5alZFFd1InRILBPHjya8Gedk/iyVpL9tILO4ik6hQ7jtd3GM7effSBtmctcmk7S+ro3RxP8ulsgejRy0Ioekxcn2a3x5JGPGTSZ+UCPHdGIt3MCij4oBCL9zGiMbxBTmbcksSkll5/ecvM6z5HzNTkTB2ol+3kN66iye+6EIc4se2JuBlz3Na7G/IegiGyJq+SiBwQ+vJ2jmGjY+GNbMvcykJ0xgUUE4U1YkMvLc/TwSERERERHxACu588dwz8JCrA0/8hnC3E3LGRvoibrOUG05qyeNZOaWBmd1ynNq+nqEPriGjTMjjnfaqMkj0TSG5SVAlxDC/cwUlFiAECauSSOh3/HuHdbP5hIzZiXFpzqmyzkUsmikiXkl9pcT1xSR0M+pzrkm7lpW0mAnA5EvpvHeuJCmTrBFnXfZidTT4gXO9nzMyyvjeajRUM2b0M538OIvX+XDMR+wa/JWSqc5/Zn8AbvGrOBPv5zMg539ODEnspH9w4vcsfI1Mveci5M5d4wjEvni6/zTCNXsqioKKdhlpqr2LBUmIiIiIiJyvvgqmacXFmL1GcKcjTv55usivinaycbZQzAczGLm3A1YPF2jG8renc7MLVYMwxLZ/N8ivinKZ3Oi45weT6a4qee9wpX26xE4mgWZRfbr8XkSE0OgeNm9JG47HrcV/Hk6y0sg6IEkvtiRzsbNO/kiaTxBlLB8+koK6ja05pB4/0qKCWFikuMa73Q6ZtYJkab9+MumM6/EgMGnkQ93rWTqshIIGU/K5446MxMZ6mMld9abpO0/k6snFwMFaw7Wr9fy2JpZLDtkc/2gjR+mq15ly/2b2XL/VOKHDiD8Cl+MDWNugy/GK3ozdOhYnr1/LZ/dn8SfrhpAaIMrbD70Ab9f8zSrv27Qzllk/jCBGJOJmR+aMeesZGq8iRjTIyzKsceH5m3H30vcUHL8XwvMGcw0mYhJyMBsySNpRjwxJhPjZqwku+LE48csyzv+Zq2Z7JQ5PGwyOfaZT1qhxem4E5j3b4Ac5k0wEWNaSW4zz8dSuIF5jlpiTI8wOyUHs+OHddmG6fb2VhW67GPNmW/f/qUsLEDuMntdy/OsFG+Ya68zfg6rv7LauzHXvzed5TnN6LvY4HwfnpXsco0Ae3fkWY846o5n6vwNFNT95izfwFSTiZj4lRQ4/+Kx5jDPZCLGNJfMC/G3rIiIiIiIAGD+z3Z7L6pxk4kPc4yR9DISHjeTBY9OZkqIgSqn7V2ee+KnM29DIZaGIZXLM4brs5G90ePPdGWlWSQ+7Hi+a6KN5VtKXNuoKSFt2UrSvmo8kIJC0t7OA8JImBlLqAHwMhB690wSrgPyVpL+ZeN7Fv/7I4qBgZOnY6rriGWMZsazcYCVpI9y6p9NrVXlAIwcGY3RMa+3cdAIRgKUW+q3s2xJJekgMG4mMwY5rnEXp2Ou23pieFmSzPOvl8CQ2cwZe2Kd5rJyjNeFMXbaeAbWDW0NjmXCWIAMChp2ZJNWR8EawJ4PeHLDAtKOur4d5DeVLfev5S3TAEI7n+YxOwcz1PQqW+5fwWt+DeZWO7qdpzbMIu1c9VyrMVOwq5CyjFe4Z9wSMgsKKdiVwbxx95I4fw6D45eQXVxCwa4Mlj9u4ukPHT9qaqvYu6uQgtJ0Xr5rDIk5xZTtKiR77VzG3Tyd9P2uxy+odLyuLWf1pKGMm5VMdk0goaEGijKWMDVmDPPymvqB3DzWz+ZyV8x0FmUUYwgNIahmO0mz4hk8aQNmIOj6SAy7Csn+U9bxf7XASvaHSyjYVUhoZCRGgMpCCnYVkvvnZ4hJSCX760IKtiUz865nmPf6BGIS1rOzvJCCbRtIHHcviwqbqgiw5rHIZD/f9HIr1JrJTJnDuJtNLK/brzyVB26OZ3bKdqyXhxBqKCZ94XRi7ppPrhUIjCCyrb2GTKdfPNacDBbtKqSgVyRRF/KcCyIiIiIirZz/lSH2YYgpS0ja5RTveIUwcto0ZkwbTpDjLfOGRxgcM51Fa3PZWwvWgg0setzE4ISM+mDImreEmJvjmZ2ylb21gHkrSbPiuWHkXHJrHBs5PdPNmzCR5R8VUlBR5dpGhmPQZHEGiRNHctfrefVBVXHyI0ydO5epc9c3PlVSRQG55YDPEPq4jIgMIWpkIFBObmHjHRUslfZEytDWtdeKwdfxemth/XDO0OtHYwDS053CtpyPSAcMt0USWlfvfzYAMDQy3GXIpyEyGhPA+lzXIaK15ax+eQ65DGHuH2Lp3cjyjv63JLIxLY25tziPSyunOA8ggkBNa9TqKVg7uJ2X17zWIFTzZex1SWy5744zn0DS2Jux961ly3X/V/9DEoCj23lszQKyD57h8U9Ddm535ny+ky92FrH5uQighOVrDaz6fCc7svP5OHEIYCVt7VbXH5o5eXSamc3/srP5oiiTOUOAgxtYk9VETy5zDh9tsUJ0Ihs3L2bBvDXsWJfIyBHRUF6C1X84c9NWMONXANHMWJHGxrTxRDbjHAr+YR8rP3FZNu/Ne5Olm7NJeTCakcF7KaoAgkcwZghQnkpmXahlzSVrHUAcpmGuX2i6JZLNn+/ki893smqcAQ5uYNFXt/Hx59ns2JlPygOBQAmLPsqjKWV/f4V5hVYMd7zJjh3pbNyYzY6k8QQdLCRxmf0Xn3l7BpkHYWBiGpuXvsmCNdlsTBzOyAFQVmIFAhn1uyFAOWu21hfOzq2pAMTffiPK1URERERELmARcSyIDYGDWcw2RdEraiTjZsxn9bYGvcSsOSxKyMBCBDPSdrJjYxqbd9ifwyypr5BUCFDO6pfmU3AwkIlJ2ezYmMbG7J2892AIlKxk6soGPQNyMrDEpfFFURHfrIrFv64Nn9Es+Gc2G9PS2PhJOnOHGSheNp3lu+y7BQ2KZeR1YZhGRzcy3RFQXkImwK8CXZ93ge7+9rgrs7i80csRes1w++cfZlFWd/61FrLTP3Ic+3hPNOOIF/jrzCEcXh3PDTebiDGN5IZxyRweNpP3Eoc7npXMFH3lOHZQg2qN/nQHoMT+3OhgXj+H2VkQOXvmac1vV/b3OSTmgeG2Cdyuac9avdYdrB3dQ9qaWSxzCdWCeXTwX3jtN8EtuLqlN6G/eZEPB9/hGh4dXctjaz6g7GhT+7WwW0bUd10NvT7yhPeCBg1nIEDt4QY7jsA0zPGDySuQocOj7f/f1Fh5r3b2//4nlTUbCinbb4WQWJYunc2M0WFndl0N9r0z1yWTXWLGYjUwcGYSC54bz8AeAP7cdIs9oEr/t6NP7q6trD4IhvuGM7BB40NvGUGQF+BlpE/faNf3MDBw2AgArE12tCsn+0NH1+fJo/Gv75Y8k807d/LFbEcg5nh/5/uppO0qx1IDoXcvZumL0zCF2YvyHzScoUBZuuNfUWoLyfq7FXziGBGttVZFRERERC5oXoGMfDWdHWsSmTgiDH9bCdlrlzAzfiR9RyaQXpc/5W51DGeczMTrDPX7xi/dyRc704gPAcpz+CgPuG48Y+qGPGIgMm48A4GyVOcRPABxTIgLqx9GWdeGf3wsQ9tasOy3YLEYuem3sUA5OQX2Ygxh41malsaCO1p+VQXjiAnMCAGyErh58ERmz5/L1DFDeaA0lKENN7aWs/PfBVgOgvUgUGvBihVLwXayS9wcFbU/g5efz8IaMo0/xDV/AQJz1hweeCoLa8h4/lof6klr1qqDNfO/FvDiAee5znwZ2+9Vnujnd1baM/abylu/HOCS9JsPLOD1f1U2uc851ZZmhV7t2p5iK//RzJk3Gn/yWPS4if+L6kOv6wZy16xUimtOvuupRD70F2ZEGClOncO4mwfS9xe96Xvz8fniAPyH3cZIoGB9DmVAQc5HWDEwdljk6Yd6p1zB1UxJDkD3Eya6NHQxYuziCM1um82C2/zhsyVMNQ2lb5/eXDtwDLP/7jSnXY8RmG4Bdq0nuxQozCH9IBh+O5wo5WoiIiIiIhcF/36xJCxNY8euIv6XuYY5sSFQksrDjydTBpi/c3QQ8Onk+vxiMGLsYrTP911RQjZAVwOdnLcJDCUcXHp71Wnndfz/69owL42nb1RU/Z8bHk4GIPvbZswzDdAl0N7ePqvL/HAAh4/YKwgPaCJ68gpjSlo6c8dFYPwpi6TkLKy/msfmxFh7WHVdYH1PtLTHx5C4xcLAxEz+l53Gxo3Z/C8zkYE/ZZF4zxTSKgCMdL/CfmhrdYOzr+88Ekj3LgBWMt+cTtrBEKa8Np5wL5ql7MME7nkgmeKQOFb9dSaRHZq3n1zcWm+wdnA7y3K3uwx5jAx+jRcHB5zVZoOGvsiyYOe+ojbScpPO6ZDQcyHojjfZ8Xk+H6clseDF8ZiCLeSmJBDjNCeAWzpEMOW9nfzv03TeW55IwrhoKLHPF1c/D1qXGxl5C5C3leyKQjJTy8EnDtNZ6fVlwBgIYIUjJ9nMKxDTgmz+l5/JxqQ3mfPAaEJ/yiPpKac57TAydMRwII+sf5sp2JZKGQbib41uwd6TIiIiIiJy7lkx78ohe1sOxU6rSBqCI4hPnM1EcDy/gKFDR/uHRxqOJHLSwWgfemkFl63M5a5ziDWhro3QJ9fwxc6dJ/55LKJ5pxUYSBTArlyKXR70rBQX5AAQFdxEbzerBYvVn1HPrWHHriK+2ZnO0plDCCrNtc+dFh1unzvNvJ20LVYgjgmxTseqW0DgYBZpOWbAQFBoGACZ/22wosC3jiDyuhCCDEDFev6SYgXMpL8wxrH4g4mn7TPxsPoZ10UewB6qPTAl1RGqzWao5lYTh1YbrJX9K8l1CKhhAi/dHn4WAgwbuR+uJrc+OPMm8vZZPOfc0NG1LPtXaYu37CnW8jyyNySz/BMLQddFYxo3kwV/nM1AwNpwsshTBVIuLBRvy2L1sg0UG0OIHBbLxBeTWPoAQInTpJhGht4yGsjio7krWFMOQQ/eRmQz/xXilFWUFDr90gghcpgByGHNFqcf3iXJ3HVVb3qNS6UMK2V5OaSlrCTTEkj4oNHEP/cmS5+NBqyk/ef4fsYbTZiAzA9fYfnfyiFwMqZm/k4TEREREZHzlYG92xIYFx/Pw2/nufYm21MXhtl7UxnDI+1TCK3OILN+xI+VzFl96HVVbxJzgJAIhvoAORlkO01hZtmebp/zzGlC/8bUtVH8jwL2Gh094boYMZRvZVNuAcWVjgprrZTtKnRdadTltCKJjjUAG1jzoVMh5ev5SwrgE0t0pOPh1xBaU5QAACAASURBVFpOwa7jUZX13/O5ISqKvg+nOs2xVs7qt5PtI44GhTnaMDh6ruVS/K1T27UljgUEwOiYMig0+jZCgbK/rSfb6dpl/20lBUDordH26+LVie7XhRF+XWCzMoCytY8QMyWV4pBYlr6jUE1cNbLmRStwtIC0/zqPOPfm0QFjCW/xmNFGbtpkHvy6CPO3+1h732QifYA2vYkfcAdvb11bn4Bn/vcDcn8zmciLIOo01BYw7/E55Pqsxzx7MkN6/ERBykqyAcN9N9q7CmOgkxEgh8SEOZSPuI2J4yJOMT7dwN4tU5j5ZyuheXuZMi4MY8VWFq0GiGDIr47/dDMOu414NpC0fgMQyIwhYS1zcp/N5YYxK7H6TOa9XdOIxMDA8bMZ+vcEMueMYVxxHJF+FnKTk8nFwNDfDiEIA2WF85k6Kw9Dmpk5D91IUE0hf1mZAxiIH+RUm/FGTOMgLWUDaUDQtCHN7pYsIiIiIiLnr/DRkxm6JIHMZWO4Yctwbr8lFEN5DqvX5mEBQh+9zT4ndHAszz6YzF3Lknl4TBXxw4KgNIuk9VYImczISMAQzYTZQ1j9VBYzTfEUTh9N2N4sVizMAEKY8eAp5v6qb2MOd91fTsJ9N9KuJJXlCzdQcDCCOZlriATMa6fwf09lwbA32bF8dCMLGBgZ+ehshn6YQGbCSGK2xzG0214yV2+gAIh8YjwjjQBmVj88lJlZMHReNqvu8McQHUdCRDKzsxK4eXAGY2/pSP6HGeRWWDEMe5MpgxyRl/FG4h8NIW1hHrPHxFMyPZYRxio+Sp1PUh4QMpn4ukXqro7j1QdTuWvZSsYNLiQ+LgLykknaZgGfOBLq5lLzH87ctOEnnE3u3N7ctQzGvpJGQj/7e5YtCcTMsI+8MhwsYNEjJhY57dPnoRUNVgyV1qZVBmvWzzfzJ+feaj6TGRvp3cKtOIVqANbV3PFnWBs3mUgjGCLH8uRna3mqrifb0bX88/MJRLZ4HR4QHMcfUyw8PWMJy5+ayHIAjESOe5MFz9UNazQy8onFjC2ezuptySzKMzJ0XMQpVgY1MPDJNOYeeYbElLlMdSwWY+gxhISkecQ7j7A1RDPiPgNJf7ZCYCxDr2uhczP6E+oDxdcHHv+lEhjL0vdg9vQ5rE5ZYu9i7BPG2NfeZM4d9q2C7l5MSmUCU5etZObElfb9ukQQ///eZM4gl4WgGTgiDkNKMlYCGTOshQJBERERERHxrMBYVm30J/GlBJK2ZJC0MMP+fpcITNOf4Q93RzielQxEPvkXUnxm8PD8DSx3THljHDSNVfMmU9cBLOi3i9jIHKbOSSVpln3YpaHHEBLeXsTEUz5GOLexkpnb7M8ohrDRzHnllfpnK0MXf3tAF9yt6aAuMJal71l5+vdzSVtv7xmGjz9DZ65g6X11iwIY6N7DfoRQx3/xCiH+L+kYXprO7HezSFoFYGTgtOUsmOy8CqmByMfWsDFoPs+/kkzSrByS7FeEyHGJ/OHJWMINTts++RdW8QhTl+WQtNB+XYwRcbz6/2Yz1I050az7zfXTGVkrCimocP28+xnOIy4XvkuOHTt2zJMFFJWU0jvk3K5Pm/v3G7mj7PjrsTd8wGu/9m3BFhqEanUMY4/3WgMs/5rFdTs+Pv550HxKf3vyaOmCU2PBYgWD0YihqZ5XNRasBiMGcyoPDEywd11uxMQ1RfX/akCtFYv9wBjP9YSRVmv96qQnqLFgsRrqFy1odPf9FqxeBoxGzZwmIiIiItIq1T3PnOq5oG67Do5FC5pg3W9/pnLr2ehUz1Yne/5puKnFgrW26ee/Jg/lqOGkz411mvHM5XxMjzwzngWeyE6keVphj7UidrokzL9h1HXnPlQDMF53M/E7Pnak7UBFAQVEOoZKXiQ6NOOHWAej/V9mOoQy5NHJTZ5/uHPvWq9m/CA9W072S6UZ52voYtRiBCIiIiIirVlzn2eaud0ZPWOcqo1mhmrgCNRO9nlTH57O811znjFP95giZ6D1BWs/fMVnPzu9bt+H0JNP7HUamh+qAWAMpk974JDj9c/5FP8A4Ze1VD0XGGME8dM0U7+IiIiIiIiIXBgugqnyT9PeUtKdX3e/2r5M8Rk7zVANgGB6d3d+vZ2yvS1SjIiIiIiIiIiInGWtLlizWm0ur0d29muBo7oTqtn5d3adU62qQX0iIiIiIiIiInJ+anXBmuWnPa5vXNLEmGtLLkkfFmA95RHdD9UA2l3iugro7p+qT9miiIiIiIiIiIh4XqsL1prFksui5GnM+nIqv087Wbh2ZqGaiIiIiIiIiIhcuFpdsGbsGOD6xrEGsZkjVHvdCmAj++umwrWWCdUOH3Md+tmzY0uuUCoiIiIiIiIiImdLqwvWDAbXoZfpBypdN/AyQBvnbezh2kMfFDmFay3XU818INfldacG9YmIiIiIiIiIyPmp1QVrdA9mpPPrvV9R5vzaJ5wp9y7gSZep12xkfjWZJzNKsbbo8M9SilxWAR1AUPemthURERERERERkfNJ6wvWLruafpc6vT6UT7GlwTY+4Uy5b8kJ4VpawQR+v6wF51SzlJJ/yOn1pX0Ivew0jyEiIiIiIiIiIh7R+oI1ehPVw/n1P9i0q5GVOJsI17IPttxCBZZdm0lyfqNHOOGnf5hmqf5yE4unxzE+Lo6pz75DbkXz9stfEce7u86w8eIUxo+Yw9Yq+0tbdTW2WqfPa21UH2p0z1Mzb+LVOZuobOSjM6+9kHfj7NfshD8rClu83rOmtoLcd55lalwc4+OeZPGmYqprT71by9RaTc6cW4h7p9hRi43qatd5BU+4H05Dk9/xl+8Q53TPueWQ+3WJiIiIiIhI69AKgzUIv+oO/J1ery7Y7DoctI5POFPi5jcI15yc0eqfe9hU8LHTa28evaqPOwc6tcK3mTQrj8ipf2Tp8j/y/Dhv3n3gWbY2JzGprmBf08uiNk8vE88vfogbOwFUkPF0AhnO6WTe24x6O8+9Y9fa2LfX1vhnZ1x7GHcu/iP/b/Ef+X+zRtGx/Shm1r0eF+beIU9W71lRydZnH+Bdr3E8v3wVSxc/SuTnCUxqTjDYIrX6Ev3QW7x8Z6j9pXkTf3jaOaxr5H44HU19x6GxvFx/z7kn/+3bWe7mbSkiIiIiIiKtg9cLL7zwgicLqNxfRVe/zue0zbaXdcT27w/IPuZ44+dcfAx3MzDA68SNDQHccE1fLi38iGzn3itnFKqBNXcl44v+R03dG21+yzOmAQRc4t7xTqbykyS+6PMYDwzww+tSb9r7X09UP186+ATSuT3syVhIbpv+XNnFvr3z673bkyjteSeXfZFCavrn7PEK4OrL7SuXVuelknMwmJ8/TeH99EIuCb6egNpiNien8M/cSowhvenaDmhTwRcpX2Lo70/pu++wPmcXpZVmvj0YQFSbnSz/awZffFVGbXklhmvD6d4OqMhj47t/Z+u/irF1v4qgLk6LOlR8yvvvrGVb8UEu72pj+6cw6NZrad/gvJus3VbM5iWfcPT6a+laNyy4eBPLsy/huqv9cL4LvNp5493OG+8j3/DPf3kz/M7r6dTOG+9L7ee/sSKYXwR4118P59eVn6WR9t5mcooPcvlVvejoDVR/yeaPbfwidDeb/5ZOTjH0Cg+kfV3EbS5m87oU/rm1sJFrHUDlphVsrAggKqS5idH/+OBlA6b5MQRf6oVXu04ERfTn6va+dA7wxasqj/fXVRDUJwBvAOfX1V+y+WP41Q1VbF5lv97dgnvRuR3Abra++zXGbrvZ9M5avqjsypVX+VG9PYXkdVsprQ3mF0GO2ou2sG1/ML/oUMD7q95nW+E37DMXY+seTvVHDe6HkE5QW01pZiqpH2yl6DtvLr864Pj1qa2m6KMU3k//nD3ewXQoTaW8Zzw3XN7gtOvvuV50PNk50uBeC7iKoE7e7MlYSErGl3z9fRV7zN5cU7dtnao83v/4IEFHPuX9v6WTf0KdFeSvSyXtH9sorQ4gKLgT3m1sFG1cQn6b/lzpB1BJfsoK/tux7u9eBTkrNlF1dTjdtYaJiIiIiIg48UR2Is3TKnus0SYc0y+cB13aWLh9NQVHm9jeGOnac+0MQzWOFpG0fa3LkNKhv7iVyLP0bfj692DXplRyvzve+8jvmv4E+9n/f19hGvn7jm/f8HX+irfY6j2Yu0y/xLb+CZ563z6O1PbNJyx/4212+Q8jJqqS5ZMmMXXBJwQMNxHNJia9tAn7INt95L9fwD686XpNGMG+vgSH3UDUVb7QqSdR13SF7mFERYXS1Rv4Lo2npq3H9ksTd43yIzdhEsvzHLUXvk3cY+kQbSIm4ifen5/Ctyc590Zr9+5J1+q32fjZ8euRv3Eh1Z16cjp5hu2bT9j+TXWjr6sznuXJ9RAVez8xPbbz5GOp7Knb8LtU3v/Ej2jTKPpUvsO417ZhA6hI46n73uDbHqO4K7Y//P0x/pBRefxaz5rHTj/HdWu2rgT0+oQP3i88PvzTtyd9InrYz/XQbrZ/spv6K9Hw9XebWJ60m6tNsQzvUcDcSQvJrwbYR/6781j0/kGiTMPw3foET056lr9V9SdmVCilCyaRtKvBdfEO4OqwYDr6BtMn6gau7NTI/UA1+QsnMbcwgOGxJvpU/ZX7p6U5erhVsvXZMbxZHMxw02Au37mQxdlNnXfdPXeKc6z8J3+Yth4GxvI7UwDbn3iM93dDx6tu4Oru0PWaG4i6JuDE++LQbraveIHFn/TgxthRdP3yFSYtzLMfs7aCjdOe4APbL4mJHUVAYSKPL8zDhjcB3hUs3uLoLVj1Oe+vSCPpk9321xWfsjYbup7O1ysiIiIiIiIe1TqDNSDo1/E86Hz21hU8t66AJkcOGiOZMvZVHjWeYaiGjdx1L/KSc0Nt7uDBXwe7e8BT8h6cQMoEXzbOGsOo2+N4bl4a+c2cYw3A1n8M8cNC8esZwZ3PTCLg7VTyHSFNt2HjuLNfTwIGxnLrFdX8X/z99OnZkz4T7ufOz4obhF7eBEREENzJj+B+/YkM9QO/UCLDesIVYUQOjCCgvY3cpBT6JMzmzn498QsdxSMvDGP76m1UYyPn/TSin7J/FhBm4pFHhrlRuzeRN48iY9sOexByaBubMwZxY7+W6yb07Zd5XD1oGL17+BIwLIHk5bEE1H3oH0v8hP4E9wwlevL9xGTk8RVADxMvrfsjE4eF4tcjjJg7+rO5sLT+mN3ufJSJoxzXreF57s4jf3cjcwXSkzvfSiTym3eYNPoW4iY9y7tbdjdvjjWAqgBufiiWPj17EDzsUWaO+pSkDMfNUxVBzEM3EdwzjJjfDiM/eBQTR4USEDqK++7uSX5Rg5usfQ/69Aula6dQogb2J9ivkfth9yYWfzeO1x+9ieAePelz/2xmdn6HjYXAN5tI+ukhXn70JoJ7hhI54VHuPtO/NqUFbO09mCHX9MC3501M+9sq7uwJvqH96XMFBIT1J7IuhGyok4n4CREE9Ajl5smT6JO1g1LAlr2KxQGTmHa3/bPoRxO49T8L2bgbfPsN4upP8uzb5W2n+u5Ygj/NoxKo/s8O9g3pf/w+ERERERERkfNeqw3W8BnAg5EDXOZayy19ilmf7GlyF7oO4InxZxKqQdknL/JYaanTO96YIuMZeAbHbA6//vfz/Kp1bFrzFhN/uZvFDzzG5maGa5FhocdftA+ld2gR3zq623m7jJ71w7vheMzTVsn35dW8l+i0UMALaXz700/YqGTfnlCCA5029+/Ble7UHjGM3336MTsPge2zj9kZczuRZ1z7cX3ufALeGYPpd08yf0Ua+c7z2Xl7Uz9tn5dzZFPNvk9SWPzsJPt5v7LF5ZjejYxUtrORn/Isj79bQKMzovmGcvOM10n+8ENWJfwW761PMGlJXuPbNhQaxpVOPaiCw/qTs7vu74i3a02+PqfV469ReyvIL3iHx+sXipjE4s+q+akG2FfJV717cjxW9CPgijNs7/pYnvdaRexdcTz30jts/M9pLNXg/D126kLXKhtWoNpcwXXXhDpdi55cff1u9uwF/CIY0OlTdlXAV58XM2C4iUh2sKvKxq4dpdw6uOcZnpCIiIiIiIicS209XYAn+f96KrOK43nsQF3EUM3qz56me/sFPNHvxF5BZ8ry2QIe++xjl4US/DtP5clft3xbzmzV1dDe1x6CePsRPOxRnq+cxPxPK7nZdOq291VVA3Xpig1bVUe6tmAIdaKeTHzjj8ScEJpUADZosKKo7SRxTpO1e4URPTyBjZ9WYsj5nOExCS15AnDFTTz9t5vgUAWl2anMjUtk4roEIk+yS/WWeTyZM5jXn3mLR3y94bOFDN7WnMa8iXzmQ7Y09pFjxVVfX/s18u4ZwZ2zZ7Dnlm189WgEp1wuw1yJcz84W/VPBLQ3QNN9O8/ckBmsfKr/ie9/BtgarCh6pmsrePXgxtnJ3Fhro/KbbfztxTjmT1jHtMFnFhE2XE3UZqsLIXsQNbia5f/ZRmVef6Kn9qDrsJ/482c78P7PYIY/d0bNioiIiIiIyDnWenusAbQJwDTmRdchoZSy8JN7eeofpS0YHdgo/scsbvlkLbku7d/BW2NuJegsfwulSXE8nlR8/I3a3eRkVdDnKkeo5uVNabFjnqeqT9ma5br/5rRN7HEEBbZPU3k3YBBRbq+26I3Bq5LqKqe3vMC7cr+jB1UPogbb+CDjeL22z95hfmoxNsdn76Ye721VmrGJnSdp7WS19x4VS37qC7xbYOLG69w4FS/4trjUPqyydjdZdXNnUU3+ikQ2fgO070HwkFHc6F/BvsZGajqxVe2j2zVhBDhCsKIdn7hRVANVW5h71xvkOF3v6k+3sfX6UHtPPy/gu2K+dXy+J2sLOS77byLjU0fhtRVkpO4gZqCbK6LWtVdVyU/1wVOD+yE8gjuzthyvt7aCzfMWkltR91kKG7+rq+2fbG00TWykzSbOsTrvHean7gYvb/xCbyJmiB97Kh3n6+VNddUpvrRG+PUbjDU11TEXHfBdGu9mDSLqWvvLgP7DKF3yBhkRN9Ab8Ivoz1dvv0X+kMH0Pu3WRERERERExJNadY81AHwG8OyYp/hhzWuk1S9eUM3qXfEU7X2K12NuJdR4Bse3FLF67VSeqmzwgN5mAG+NmXrWh4AC9I5/g5tfeoJRt/tyuS/8WFlNVPwbzHSESX3ufIKuT0xiWBL4hj/ExIHgPFj17iHeLI+Lo9TLxo+14Ux/YxS+wGkMmnPiR3RsBI9Pu52tdyeyND4Mrh/FI28/hun27cz8cwI3xiZy90tPYLoLurWva9M+tC4gNpH7np1E7F2+dMObqyebuPuzpltrqnYAeg3m1kPvsDnmCdyZqstv2CTu3PQEo0aDX6dRTBxeFzj5cnX/AFZMu521vr5wyEbw3W8w0x97p7umjjf8fsInTiIuzRdDrTcD+oc3vXGzixzF9GcKefZ3t7DYzw/DoUq+9x3G83XXwX8YE02beOz2W6CTH3fGDyPaef8wE8E7HmP8QhtUV9PRlMhr7oSQdfwHc8f1jzHprk+4L3EVd4c1vB8G8cirRTx13+2s8PU93mYPgEE88ocCHn/Efl2tvsOYOC6M/FO22fQ5+vbuT8Dbj2F635du2LAGx/LybHvg3GfU/ax4bAymnCdIee4mmr2mQM9Ynp8wh+fixoCvNz9W+/K7V984PtS412Bu9X2H0l86vt9eEUTbUvjphjMILEVERERERMQjLjl27NgxTxZQVFJK75CzN3F/c1m/XsuTGxY4hWsObfwwXfk0j904gNDTWdn2QCmZW5fw0rfbKT7hmAN4bfSLjL2q5SbLbxZbNdWHwNvX9yTzdTWhwZDCc6LWRrXNG9/Ghp0eqsbm3czzaLL23bz/wBt4J75FTI8WqLcRtqpqOM3r7TJ0t4VrsXn7Nn49T+V0rncLafI6NOdePLSN+b8rIGbdQ83rBXaommovX1r09q61UV0Nvp3O8d9zERERERG56Jwv2YmcSMGasz0f83Laiyw71NjETd6Edr6V+64cQNTVfQjq5ovR4PSxtRrLj3vY+VUu2d++y9oDlZgbOYp/+1t53fQUQ7X0n0dVF24ja0syi6viSHtu0JlPui/nj0O72bowkbm1caQ9o+9WREREREQufOdVdiIuFKw19PMe0lNn8dwPRY0GY+7zZuBlT/Na7G8IurRFDyxuqC7+lK9+CuDq63view57Yck5cKiC/LxKLu8fhp++WxERERERuQicd9mJ1FOw1gRL8Qcs/OcSlh08/cnLGwryuYMnb5qAKbTZszSJiIiIiIiIiADnb3YiCtZOyfLtdtZtS2KZuYCy09rTl6H+8dw36FaGXqlATURERERERETcc75nJ62ZVgU9BeOVA4i/cgDxP9swf5dPdmEu+ebt5B+CvdYi+8IEbfwYaOgK7ftwk39vwsOiibrCD4OGfIqIiIiIiIiIXLTUY01ERERERERE5Dym7OT81cbTBYiIiIiIiIiIiFyIFKyJiIiIiIiIiIi4QcGaiIiIiIiIiIiIGxSsiYiIiIiIiIiIuEHBmoiIiIiIiIiIiBsUrImIiIiIiIiIiLhBwZqIiIiIiIiIiIgbFKyJiIiIiIiIiIi4QcGaiIiIiIiIiIiIGxSsiYiIiIiIiIiIuEHBmoiIiIiIiIiIiBsUrImIiIiIiIiIiLhBwZqIiIiIiIiIiIgbFKyJiIiIiIiIiIi4QcGaiIiIiIiIiIiIGxSsiYiIiIiIiIiIuEHBmoiIiIiIiIiIiBsUrImIiIiIiIiIiLhBwZqIiIiIiIiIiIgbFKyJiIiIiIiIiIi4QcGaiIiIiIiIiIiIGxSsiYiIiIiIiIiIuEHBmoiIiIiIiIiIiBsUrImIiIiIiIiIiLhBwZqIiIiIiIiIiIgbFKyJiIiIiIiIiIi4QcGaiIiIiIiIiIiIG9p6uoDzUU3NISr3H6DKUs3BmhoOW23UHjnCMU8XJiIXhEsAr7ZtaWfwxqdDBzoZffHr0pkOHdp7ujQRERERERFpQQrWnOz5wcz3FT9w+LCVrn5d6OrXmeCgy2nXzkDbtrpUItJ8R44c4fBhK9UHazhg+YnS8u9p187A5T0uI+Ayf0+XJyIiIiIiIi1AaRFg/rGSb0rLMHh7Exx4Od26+nm6JBG5wLVt2xZf37b4+vrQwxGk/bivku++/4Gy8u/pFRyEfzf9rBEREREREbmQtfpg7cuirzlQZSGkV08FaiJyVnXr6ke3rn78uK+Skm92U7n/ANf0vsrTZYmIiIiIiIibWu3iBVabjZ1f5HMM6N/vlwrVROSc6dbVj/79fskxYOcX+VhtNk+XJCIiIiIiIm5olcGa1Wbji13/pWuXzlyr3iIi4iHX9r6Krl0688Wu/ypcExERERERuQC1ymAt/79f0d2/K1f2DPR0KSLSyl3ZM5Du/l3J/+9Xni5FRERERERETlOrC9a+LPoanw4dFKqJyHnjyp6B+HTowJdFX3u6FBERERERETkNrSpYM/9YyYEqi4Z/ish559reV3GgyoL5x0pPlyIiIiIiIiLN1KqCtW9Kywjp1dPTZYiINCqkV0++KS3zdBkiIiIiIiLSTK0mWNvzgxmDt7dW/xSR81a3rn4YvL3Z84PZ06WIiIiIiIhIM7SaYO37ih+44vLLPF2GiMhJXXH5ZXxf8YOnyxAREREREZFmaBXBWk3NIQ4ftp693moVqTxwVW/GvVve8se2WrDst2CpaflDi0ON4xpbz2IbtVZ7G2faSEsdp6WcB/WUfTiHcQ+vpOAkf0esu1bycPwc0s/CX9GW1q2rH4cPW6mpOeTpUkREREREROQUWkWwVrn/AF39uni6DLeYN8ygb1QUfd/K83QpLixf5ZC9LYfc8vMk4DkDuW9F0TcqiqkbzuLwu7z59u9x6nrOqJWWOk5L8Xg9ZnLXJpP90XzSv7S/09i9WfCP+aRvS2bNv8+Lq3ZKXf26ULn/gKfLEBERERERkVNoFcFalaWazsaOni7jolL8fjzj4uNZtN3i6VKkVfPHNC+d99IymRFhf6exezPysUw2rklnwW3+Hqrz9HQ2dqTKUu3pMkREREREROQU2nq6gHPhYE0NwUGXe7qMFlH20XzWFIL/4PHE9zMCFnJTVpK5F8LvnMbI4LrX/gz9/QgO///27j8sqvvA9/gnzsDBAQ+igCaBjQbS3IJpHnBz78WbPotXN3iTDXTTYn4ZdxM1bdRuYtzNtT67de226rZr3LZq7tZoW8Rs1U1b1KaYbC6apJJqgGQjbFJAk0jTyO8ZBDkww9k/BhQNyDiiB8L79Tw+zzDMnO/nfM/hDz/P95zz0lbtPOhVav4yLb83Rb7SQm0u2Kv6+Hw98lcLNCtBkq9CBdsPqUHpyluUpMqtW7Ttw0TNf3ipFt756SLi1MFNKjgWfF1zcKs2Wrla8nCGTEny1apk/14VHSyVdVOusvNydf8ff3ob1u/2a9uvamRNytaSv+j9bqBWxT/Yp0qZynpwkWZNlfRJhXbv26dDv6qVccddynswV7NTzOBGPnxZG39eKSVmnxvf91ahtr3eIKXla2VO0rn5Ss9dIPPV9dryRqbWFixQ6lATfdG48xcvCObpE2jQkZ8V6uAbh1TjyVJ2dq7uvztNpqv/77dq5556JT6yWCtvHGCMgE81h/Zpz0sv60hHivLuzj+/jb5jkpit+XfUaeem52V9ZafWxp7/eucnpSrYUqiDHanKXbBA92ckDL7tObn60p9nKMF1/iO+qv3a/fMSHTpmKTU7SzkP9u1jv3MoP0unCp/VNitfL6zJlhnKfvXpd3wWzunUL7/3nA4pQ/MXLVJemnlZWRveKFTBwVKV/Ic0855s5eXmK7Mv6/59KqmX6s0VSn9v4HPT+++FKq6SEtT3d/PpYzzguZWWr+X/rUa7n9+rgx3JyvnKIi3MeHV6iwAAFgVJREFUuvrlXEy0Rx/VfXzVxwEAAAAAXCHbYb+r+eCqj3H4N0ft7u7uqzfAH/bYj05PtR/611PDvun6vYvtadNT7Wnry23btu2y9an2tOmp9qN76/s+Yf/s0eB73znW/+dH7CefzAl+d3qqPW16uv3oMyvsL577OdWeNneLfdx/Pv+06fn2ffelXvCdhwprPpWpL8O5f4/usett27abS+xvzg2+d2tWjn3HjODruevL7c6LN9J5xP7mjFR72vQV9q+9ve9Vbgnmy91pV9u2bdfstO/r3cYX5ubYX+jbj72983xs/YXjX2K+vrZsaW/e9XbZIPtzbk5P7bEfnREc6457+sbNsb9zrHcvOsvtf5rblyvXvicrPbjPTx60vbZt2/5T9s8Wp18wR3Pn5lyUtdMuW9/7XmbO+W0s3mN/1P+YPLTU/toX++Xr2+fcxfajX7zwWD35q76J9Nr//297tz0jy57bu+1pc9fbZe298/Tzxfatvb+/J/f8Pv6w8sJz6GvLss/nDmm/+jmXNf/ccQz+y7S/+XpnyFmPb+03T31ZZyy2f/mHT5//g52bnzrGoZ5by1bYT16QPcf+p/JPnc3Drru72z78m6NXfRwAAAAAo8O16E4QnjFxKWjA75fbPUyL8zoqtPHPZuqxfxviLuh1L2vVn87WujeculSyVKdS1uud6mq99g/ZkiyVlCZqw9vVOvn2Ni2MllS7VyXv9/9Og2auKNPJE9V6p2CRkmXpyN8VquSi26hlrqrWi48HX8/+7hGd3JGvBEnl/7JcBbVS6uN79M6RYh39TYGWJEk1P/qGCn53UTwjUzlfMSTtV2l5cICaYwd1SlLm/GylqkG7v7NW5e2GZn+3RO+8Uqx3StZotiyVrN2q4suc1uL3k7XryHG9d3yFMi/5SZ+KNqxWSXuKluwp09EDxXrntXWaHV2rbev36pQkq7ZOunupln+3SEdfKdKB17dpiSRr316VNEhW6XatedWSkhZp19vVOlldprV/ctEkvrtdT/2oVsacdXrtaLEOHCnTi4+nyHp1rZ4v7ffZ0sOylhbrnePH9dy9/VZKvXtKqc8e18kT1XptXbYMWSraUKhKSXprq57YVSulLNKLvz2iV14/ol2PJUm12/V/C2slNai6JU1Lvr5KO359RAeKinVgXZakWm0+2P9efqUqsZbqlbeP673NuTJD2a+BvBurvAPVwaz/kC3Jp4Lv7VVNSFkrVLy1VlKG1v68WAeKivX6njVa/liavHWfPgkGOzcvdBnn1uE2Zf3yuE5Wl2nHY0m9c1Q19D5fIbfbrYDff9XHAQAAAABcmTFRrA0rf6csy6eSZ+YNXq7VvaxVjy7T7lpDcg38kWthZlaGTJeUfGtK8I3UFN1iSjLTlXaHJNVd9LTRHM27M3gpnHlnruYnSdIhVdaGMlqVjrxkSUrTI/kZMiTJzNK8uyWpVqX/cfFN4w3NumeBDEm736yS1KCyw1WSMpR3Z5Lkq1DpIUnK1yP3JgW/clOOcrIlte9V+cVF3RBmf3WRZk01ZHiMS3/Q96ZKXpIkU/Wvb9XGTZu0cW+VFCep4rDKGyQj7V6tXLpAedO8Kt61SRv//jnt7v16Z0A6VVsuS1Lyg7maZUpymZp1T676j1z5RrCkS1SV9vxgkzZu2qqSBlOSpd3H+hc3C7T8gRSZHkNG/w1kLdIjfxx8I/nPH9b9klRXoZoGqfLYQVmS0hfkK9PTO/68HElSzZvlalCCZj22Qsvvy5Lx/l5t27RBGwvLJUnWRT3Zwq/mK9UMzlso+zWg7Lv0f24Kvky+N195kvRuuWp8oWRNUfrdhqQKrXtwvp769nb92pup+UtXnL+k83Jdzrl1x1363ymG5DI1e05OeOMBAAAAAD6zxsQ91lxut/zDtWrNzNLqoj1S3nxte2aeHlOxdtzZ7/d1+/VE3tMqbknRkl17tDorzP/8O85UQrKkuovLt8FY8tVJUqKM6BCHyMjV8qTt2vhSqSqXpgbLjoxczb5J0ideBRcOGYodsrkZRh1949aq8pAVXFUlSXFpSo9LCJZItYX68peCK54S0lI0MyNFqZLKez/qaw42kWb/Es9MUKoUXFEmyfIGS9n6ynKV1J//WPptaVIo82dIUedex54rtzoD57edGD3YuedTyd/N12O7aqW4FKWnp2nG9QlS1aVXYYayX0OKnqBESVKbvB1S4pBZTc1bU6R/nrpFe146rKIdG1S0Q1J0mlYW7NHyjFAH7qcjzHMrIoyxwuT3++UarlW2AAAAAICrZkysWIsyItXZGcIla6HyZGh10R4tSbGCK9d+UitLUuf7hVdWqjW8rFV5efqzvE060hu3/g/BamfWtAsvaKv39V6vFvDJ13SF+3NOuY5/2PuypUKHSiUpW+k3hfLdFGXmStIhVVT2zXWdjvfeSD45foC5cKVpdn6SVHdQJT8tVZGk2Q/mKFmSpqYrM0mSSlXWt4LIqlHVMUlKU1Jcv+00+eQNBF96ff1aqnBMTVV6tCTlaPW/FulAUZEOFO3RC/+yUy/8ZJVmT5LK921Qebt0/7YyHT1QpOf+Pl8z+20ieVqWJKnyWJX6ripsqDh8QfmUfGu2JCl18bO9YxTpwM7n9eOf7NQLC0Joi45V6Xhf4fm7ch2RJKUoOUFK/cK9kqSSY5XqOxKn3u+t/W5MkNlwWDt31Uq3rdIrR4t1oOBZLZuTPOSQoezXYFmr+4JUVahYkqLTdMvUELIGLPmsBGXOXabnXinTyeoy7fp6mtRepY3hXpJ5OeeWQzo7LUUZkU7HAAAAAAAMYUwsiYj2eHSmvUMxMaEupQpBb7mmvPna9qPtwfd+ul3SFaxUS0hTmrtKuyuq9MQSnxZ+rl7FO+okZSgnK3jJWrCI2K/Kb8/Xlyv/RLGlVaoZtqNYpXXLlqluTrLqXy1UsSQjN1+zB3gIohGbJKlOR/Zs0kYrX0seztDs+xcped927X7qEZlr8pXwRqE2VkhKWaH5Xxx4aVB6dr6SN23Sxk1Vku5S3py+wdKUtzRbG1cf0rply6SvZqth3wYVtEtG7jJ9KUWSL1N52q6idzfoy/MrNXvSm6p8/0qXt2Uo768ztHntXj3xl6bWfvV/SqVbtWZHhayMNXrtxQUyohIk1ansjX06YiTp1KENKui3hYTsfOVFH1LRS9/QQ09VanZ8vUp+USpDOlceJeQ8rIXfPKSCby/TE/5leuT6eu38wQYV1xpaWFCmtUM+tnSvnprv1f1zYlVZGLy3WvLjOZrpkozsfC1J2q9te5froQlrND+xVDt/WCEpRSsfzJZhvBx8CuuJN1V8KE2ZqtLO75UOOTOh7NdgWZ94xNLCLEPlhYU6JSl1aU7wXndDZVWVts3P08ZaQ5lfX6+Vd0ilpcGVc5nJAz+dc6Bz80IhnFtvDTkdV9WZ9g5FezzOhgAAAAAADGlMrFiLNWPU6msb/g2fW7nW98aVXv6ZpIXf36KFGaZ8bxRq846XVROXpZW7tmhh76oxM+dpPfdAmgz5VP6Lw7L+8lmtvXsY9kWStEBrl8bqyI7tKqqyZN65Si+uu0sD7U36A+u0JM2Q9dZebd5wSDWSjKxV2rV5gTIjK7TtmdVat69Kxp0rtOuFpUof7F5zt2X33stN0t3zNLvfaqHkBzbrwKpsJXzystY9s1rb3rCU/sA6HejLZN6llZvzlR4t+Sr2q8RapH9ec+X3wUpd8LxeXJUt8z+3a9WSJVq1o0rmnFV68ccLlCwp/S+e1coMUzU/Xa2HFy7RzugFWpndbwNxd+kfX1ilWXGWKvdt1+ZfWFqyeY1m9R/Ek621B57VwowGFW94Wg8/uUHFzRla+P1irb0zhHLwjlX6x/satfuHhTrSYighd412/FXvve08WVr9whYtzDBUvmO1Vm3Yr8rILK3ctVPL03rnbeO9StAhbVyyUA8/U655f7No6DFD2a8Bs67Qt+bWqaA3a/oDW7Tj8bTeeRgiqytNy3/c+/sfPq2HFz6tzf9pavaKAv2/B5IGHG6gc/NiQ55bDmv1tSnWjHE6BgAAAABgCNfZtm07GaC69kPdkhLStYZh6+g4q4p3q/S//sfMoT8c1gAVWpf3Delbw3hPNcsnn2XINK/BDcY+2avHZq1WiRbpxROrlBmwgmOHsmCmw5IGeCCA1eKT5TE1dHyfipbP1FMvSfO+X6bn7h1g/gKWfD5LhmnKuJYPg+gdV4PtR4dPlnHpTFaHNfQDEzp6j3VcGMc6YMnnNwaf5w5f8PcDfSBgydeusM6xkPbrrQ2aPn+7lL1OR3fkK+FKskrBv4kOhX4eDHJuXsCpc2sIv/ltmTJuS5PHM97pKAAAAABGgGvRnSA8Y+JSUI9nvKKiDDU2NSt+8qSrMECGVr9SPLzbNEIppa4SV4ilmjRocWHEmUM+LbLhF0/r0a2Hg08dTVmqJXcNUkq6wiydrtRQ43qG3schy6fe7YQ83xdzGTIvVQh5zMFXYLkMmWH2wCHt10DjhZtVuvy/iVAyOnVuXUJjU7OiogxKNQAAAAAYBcZEsSZJN0ydot9/fPrqFGujnStWibelKV0D37Pq6rBU/UGt5ElSek6OVn9rqTJHVr+BK+VKCD7ldGqs00lGld9/fFo3TJ3idAwAAAAAQAjGxKWgfY6WvaObpyVTrgEYkRqbmnXig1P67zNvdzoKAAAAgBGES0FHrjHx8II+029KVu3Jj5yOAQADqj35kabflOx0DAAAAABAiMZUsZYQP0kTY029V33C6SgAcIH3qk9oYqyphHhW1AIAAADAaDGmijVJuvWWm9Xe0aEPPqpzOgoASJI++KhO7R0duvWWm52OAgAAAAC4DGOuWJOkGZ//nOobmijXADjug4/qVN/QpBmf/5zTUQAAAAAAl2lMFmtGZKRuv+3zampp5bJQAI55r/qEmlpadfttn5cRGel0HAAAAADAZRqTxZoULNdm3j5D10n67Vtvq7Gp2elIAMaIxqZm/fatt3WdpJm3z6BUAwAAAIBRyu10AKfdesvNamhs1okPTun3H5/WjTdMUfxkbh4OYPg1NjXr9x+fltXVpZun/REPKgAAAACAUW7MF2tS8GmhCfGT9IfTDfqw7mO9X3NSkyfFaaI5QTHRHkVFGXK7mSoAofP7/erstHSmvUOtvjY1NbcoKsrQDVOn6PopCU7HAwAAAAAMA9qifq6fkqDrpySoo+Osmlta1dTcqo/qPlan1aWA3y/b6YAARoXrJLncbkUZkYr2eBRrxuiPbrxeHs94p6MBAAAAAIYRxdoAPJ7x8njGK+lGp5MAAAAAAABgpBqzDy8AAAAAAAAArgTFGgAAAAAAABAGijUAAAAAAAAgDBRrAAAAAAAAQBgo1gAAAAAAAIAwUKwBAAAAAAAAYaBYAwAAAAAAAMJAsQYAAAAAAACEgWINAAAAAAAACIPjxZrLNU6BQI/TMQAAAAAAAEacQKBHLpfj9Q0G4fiRcbvd6vZ3Ox0DAAAAAABgxOn2d8vtdjsdA4NwvFgzjEhZnV1OxwAAAAAAABhxrM4uGUak0zEwCMeLNU9UlDrOdjodAwAAAAAAYMTpONspT1SU0zEwCMeLtejo8TrT3sF91gAAAAAAAPoJBHp0pr1D0dHjnY6CQTherI0bN06mGSOvr83pKAAAAAAAACOG19cm04zRuHGO1zcYxIg4MhNjTbW0+li1BgAAAAAAoOBqtZZWnybGmk5HwSWMiGLNiIxQrBmjxqYWp6MAAAAAAAA4rrGpRbFmjIzICKej4BJGRLEmSfGT42R1dcnr5ZJQAAAAAAAwdnm9bbK6uhQ/Oc7pKBjCiCnWJGlKYryaWlrVdqbd6SgAAAAAAADXXNuZdjW1tGpKYrzTURCCEVWsGZERun5qohoam1m5BgAAAAAAxhSvt00Njc26fmoil4COEtfZtm07HeJiVle3Ttc3yoiMVPzkOLlcI6r/AwAAAAAAGDaBQI8am1pkdXVpSmI8pdooMiKLtT6NTS3y+s4obqKpWHMCBRsAAAAAAPjMCAR65PW1qaXVp1gzhnuqjUIjuliTgqvXWr0++XxnFBPtkWd8lIyoSEW4IyjaAAAAAADAqBEI9Kjb3y2rs0sdZzt1pr1DphmjibEmq9RGqRFfrPXp6elRe/tZdXR2yrK65Pf7FQj0OB0LAAAAAAAgJC7XOLndbhlGpDxRUYqOHq9x41g0NJqNmmINAAAAAAAAGEmoRQEAAAAAAIAwUKwBAAAAAAAAYaBYAwAAAAAAAMJAsQYAAAAAAACEgWINAAAAAAAACAPFGgAAAAAAABAGijUAAAAAAAAgDBRrAAAAAAAAQBgo1gAAAAAAAIAwUKwBAAAAAAAAYaBYAwAAAAAAAMJAsQYAAAAAAACEgWINAAAAAAAACAPFGgAAAAAAABAGijUAAAAAAAAgDBRrAAAAAAAAQBgo1gAAAAAAAIAwUKwBAAAAAAAAYaBYAwAAAAAAAMJAsQYAAAAAAACEgWINAAAAAAAACAPFGgAAAAAAABAGijUAAAAAAAAgDG6nAwwHv9+vjrOWLMtSV1e3uv1+BQIB9fTYTkcDAAAAAABjxLhx18nlcinC7VZkZIQMw5BnvCG3+zNRv2AA19m2PSrbp0CgR762MzrT3qHu7m6NHx+l8YahSCNSEW63XK5xGjeOBXkAAAAAAODa6OnpUSDQo26/X11Wl85als6e7VRERIRioj0yJ8TI5aKr+CwZdcWa3+9XS6tPrd42mRNiNGFCtDzjo5yOBQAAAAAAMKCOs51qa2uXr+2MJsZOUNxEk1VsnxGjqlhrbvGqqblVcRNjFTdxglwul9ORAAAAAAAAQhIIBNTS2qaWVq8mT5qoSXGxTkfCFRoVxZpldam+sVlut0vxk+IUEUGrCwAAAAAARqfubr8am1vk9weUGD9JhhHpdCSEacQXa762dp2ub1RiwmTFmjFOxwEAAAAAABgWXt8Z1Tc0aUpivMwJ0U7HQRhGdLHW0uqT19umqVPiFRVlOB0HAAAAAABgWHV2WvrkdKNie++9htFlxBZrLa0++drO6IapiVz6CQAAAAAAPrO6u/36+JN6mRNiKNdGmRH5jFdfW7u83jZKNQAAAAAA8JkXEeHWDVMT5fW2ydfW7nQcXIYRV6xZVpdO1zdq6pR4SjUAAAAAADAmRES4NXVKvE7XN8qyupyOgxCNuGKtvrFZiQmTuacaAAAAAAAYU6KiDCUmTFZ9Y7PTURCiEVWsNbd45Xa7ePonAAAAAAAYk2LNGLndLjW3eJ2OghCMmGLN7/erqblV8ZPinI4CAAAAAADgmPhJcWpqbpXf73c6CoYwYoq1llaf4ibGcl81AAAAAAAwpkVEuBU3MVYtrT6no2AII6JYCwR61OptU9zECU5HAQAAAAAAcFzcxAlq9bYpEOhxOgouYUQsD/O1nZE5IUYul+vaDNjj04ev/ky7i4+r2tutbkkRhqlbbvtT3f/gXN0Ue21iAAAAAAAADMTlcsmcECNf2xnFTTSdjoNB/BcpqVu6GPpgdgAAAABJRU5ErkJggg=="
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![image.png](attachment:image.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "OcYcfQ3xi356"
   },
   "source": [
    "### Submitting the above result in Kaggle gives me a score of 0.98042"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "LqJ-ocPOnTnQ"
   },
   "source": [
    "# CNN Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "C7Xok05y4Yp1"
   },
   "source": [
    "From face recognition on a phone camera, to self-driving algorithms employed in the most high-end electronic vehicles on the market today - CNN or Convolutional Neural Networks are at the base of most of the breakthrough innovations happening today. CNNs are a type of ANN or Atrificial Neural Network that are mostly used in image processing.\n",
    "\n",
    "How it works:\n",
    "Pixels - we've all heard of this term. With camera phones flexing their dozens or even hundreds of mega-pixel cameras, the term pixel has found a place in common vocabulary in the past few years. A pixel is nothing but a unit area of an image. It is generally represented by a square of 1x1 size.\n",
    "Essnetially, an image can be represented as a matrix of pixels. The more the pixel, the more information can be captured within the picture. Mostly, each pixel contains information in 3 dimensions or channels (most used - Red, Green and Blue channels). These channels talk about how intense a particular color is for a particular pixel.\n",
    "\n",
    "Now, in these matrix of pixels - there are a lot of patterns. For purposes such as image recognition or classification, we can use these patterns to identify which part of an image represents what. For example, in recognizing handwriting, as we are doing here, we can focus on some specific strokes or curves as features. If there is a curve present in an image - it'd be one of the digits that have a curve, say 2,3,5,6,8,9, or 0. Now based on how many curves (features) and their relative locations, we can differentiate between numbers.\n",
    "\n",
    "That is exactly what a CNN does - it adds some hidden layers - whose purpose is to derive features from images that are later used for classification, identification, and recognition."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "id": "Tvr60pEkoCup"
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Conv2D, Lambda, MaxPooling2D, Dense, Dropout, Flatten # core layers and convulation layers\n",
    "from keras.datasets import mnist\n",
    "from keras.layers import Dense, Dropout, Flatten\n",
    "from tensorflow.keras.layers import Conv2D, MaxPooling2D\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import (BatchNormalization, SeparableConv2D, MaxPooling2D, Activation, Flatten, Dropout, Dense)\n",
    "from tensorflow.keras import backend as K\n",
    "\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from keras.utils.np_utils import to_categorical"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "id": "ZhF0uQXSnRYf"
   },
   "outputs": [],
   "source": [
    "#creating a new dataframe\n",
    "digit1 = digit"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Vd47awFBnvti"
   },
   "source": [
    "Set data features and labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "id": "vOz_8IvcnhyC"
   },
   "outputs": [],
   "source": [
    "X = digit1.drop(['label'], 1).values\n",
    "y = digit1['label'].values"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "CrIuB1LqnxL-"
   },
   "source": [
    "Reshape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "id": "Tm9IsD92nh0z"
   },
   "outputs": [],
   "source": [
    "#Scaling the data \n",
    "''' dividing all the values by 255 will convert it to range from 0 to 1'''\n",
    "X = X / 255.0 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "id": "naZCwe5ctj-o"
   },
   "outputs": [],
   "source": [
    "# Reshape image in 3 dimensions (height = 28px, width = 28px , canal = 1)\n",
    "# canal = 1 => For gray scale\n",
    "X = X.reshape(-1,28,28,1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Aloqjovxn2PT"
   },
   "source": [
    "## One-Hot Encoding\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "l_vlTc3snh3o",
    "outputId": "00929041-913a-4941-bf7c-0fdda1904469"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Label size (42000, 10)\n"
     ]
    }
   ],
   "source": [
    "# Encode labels to vectors\n",
    "y = to_categorical(y)\n",
    "\n",
    "print(f\"Label size {y.shape}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "HwRCosKFsZ9P"
   },
   "source": [
    "## Split training and valdiation set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "id": "yonQklFpsaiO"
   },
   "outputs": [],
   "source": [
    "# Split the train and the validation set for the fitting\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.1, random_state=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "F6z0ridSsdX2",
    "outputId": "3f5aa2e7-2aed-4fdb-ec4c-c4b637291b1a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((37800, 28, 28, 1), (4200, 28, 28, 1), (37800, 10), (4200, 10))"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape, X_test.shape, y_train.shape, y_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "eO196MrbslK8"
   },
   "source": [
    "## Normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "id": "jD_Nb2w_sltq"
   },
   "outputs": [],
   "source": [
    "mean = np.mean(X_train)\n",
    "std = np.std(X_train)\n",
    "\n",
    "def standardize(x):\n",
    "    return (x-mean)/std"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 307
    },
    "id": "sWAiNMzRtT5X",
    "outputId": "532acf56-b8a7-429a-9b3b-dae41c6a8c49"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "X_train__ = X_train.reshape(X_train.shape[0], 28, 28)\n",
    "\n",
    "fig, axis = plt.subplots(1, 4, figsize=(20, 10))\n",
    "for i, ax in enumerate(axis.flat):\n",
    "    ax.imshow(X_train__[i], cmap='binary')\n",
    "    digit = y_train[i].argmax()\n",
    "    ax.set(title = f\"Real Number is {digit}\");"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "x-_xsYMgrXjP"
   },
   "source": [
    "## Building Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "id": "dnOScHD7nh6e"
   },
   "outputs": [],
   "source": [
    "model=Sequential()\n",
    " \n",
    "model.add(Conv2D(filters=64, kernel_size = (3,3), activation=\"relu\", input_shape=(28,28,1)))\n",
    "model.add(Conv2D(filters=64, kernel_size = (3,3), activation=\"relu\"))\n",
    "model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "model.add(BatchNormalization())\n",
    "model.add(Conv2D(filters=128, kernel_size = (3,3), activation=\"relu\"))\n",
    "model.add(Conv2D(filters=128, kernel_size = (3,3), activation=\"relu\"))\n",
    "model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "model.add(BatchNormalization())    \n",
    "model.add(Conv2D(filters=256, kernel_size = (3,3), activation=\"relu\"))\n",
    "model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "model.add(BatchNormalization())\n",
    "model.add(Flatten())\n",
    "model.add(Dense(512,activation=\"relu\"))  \n",
    "model.add(Dense(10,activation=\"softmax\"))    \n",
    "model.compile(loss=\"categorical_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "NV_wOcKfnh88",
    "outputId": "a683f579-9621-4e81-fece-e195e278fc5c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 26, 26, 64)        640       \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 24, 24, 64)        36928     \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 12, 12, 64)        0         \n",
      "_________________________________________________________________\n",
      "batch_normalization (BatchNo (None, 12, 12, 64)        256       \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 10, 10, 128)       73856     \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 8, 8, 128)         147584    \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 4, 4, 128)         0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 4, 4, 128)         512       \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 2, 2, 256)         295168    \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 1, 1, 256)         0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 1, 1, 256)         1024      \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 512)               131584    \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 10)                5130      \n",
      "=================================================================\n",
      "Total params: 692,682\n",
      "Trainable params: 691,786\n",
      "Non-trainable params: 896\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "hZHswJntsG-8"
   },
   "source": [
    "Data augmentation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vbcUO7ctsLid"
   },
   "source": [
    "Data augmentation is a technique in which one alters the training data that change representation, while keeping the label same. This technique is one of the ways to avoid the problem of overfitting. In this solution, we can expand our handwritten digit dataset. We can make the existing dataset even larger. The idea is to alter the training data with small transformations to reproduce the variations occuring when someone is writing a digit.\n",
    "\n",
    "By applying just a couple of these transformations to our training data, we can easily double or triple the number of training examples. This massive increase in training data leads to much more robust results and better models."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "-wmJ4YZVnh_g"
   },
   "outputs": [],
   "source": [
    "datagen = ImageDataGenerator(\n",
    "        featurewise_center=False,  #set input mean to 0 over the dataset\n",
    "        samplewise_center=False,  #set each sample mean to 0\n",
    "        featurewise_std_normalization=False,  #divide inputs by std\n",
    "        samplewise_std_normalization=False,  #divide each input by its std\n",
    "        zca_whitening=False,  #ZCA whitening\n",
    "        rotation_range=15,  #rotate images\n",
    "        zoom_range = 0.015, #Randomly zoom images \n",
    "        width_shift_range=0.15,  #shift images horizontally\n",
    "        height_shift_range=0.15,  #shift images vertically\n",
    "        horizontal_flip=False,  #flip images horizontally\n",
    "        vertical_flip=False)  #flip images vertically\n",
    "\n",
    "\n",
    "#datagen.fit(X_train)\n",
    "train_gen = datagen.flow(X_train, y_train, batch_size=128)\n",
    "test_gen = datagen.flow(X_test, y_test, batch_size=128)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "BTy1nxJ9trGL"
   },
   "source": [
    "Data was augmented (to enhance the training set and accuracy, as described) by:\n",
    "\n",
    "* Rotating randomized training images by 15 degrees\n",
    "* Zooming random training images by 15% \n",
    "* Horizontally and Vertically shifting images by 15%"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "b9Di_nWotx0-"
   },
   "source": [
    "## Model training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "id": "64z681qIsH35"
   },
   "outputs": [],
   "source": [
    "epochs = 3\n",
    "batch_size = 128\n",
    "train_steps = X_train.shape[0] // batch_size\n",
    "valid_steps = X_test.shape[0] // batch_size\n",
    "\n",
    "es = keras.callbacks.EarlyStopping(\n",
    "        monitor=\"val_acc\", # metrics to monitor\n",
    "        patience=10, # how many epochs before stop\n",
    "        verbose=1,\n",
    "        mode=\"max\", # we need the maximum accuracy.\n",
    "        restore_best_weights=True, # \n",
    "     )\n",
    "\n",
    "rp = keras.callbacks.ReduceLROnPlateau(\n",
    "        monitor=\"val_acc\",\n",
    "        factor=0.2,\n",
    "        patience=3,\n",
    "        verbose=1,\n",
    "        mode=\"max\",\n",
    "        min_lr=0.00001,\n",
    "     )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "f4X3NJcesH6c",
    "outputId": "a188f701-0144-4c30-d9c3-badce1f6887a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From <ipython-input-18-2676189ccf2f>:2: Model.fit_generator (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use Model.fit, which supports generators.\n",
      "Epoch 1/3\n",
      "295/295 [==============================] - ETA: 0s - loss: 0.2001 - accuracy: 0.9368WARNING:tensorflow:Early stopping conditioned on metric `val_acc` which is not available. Available metrics are: loss,accuracy,val_loss,val_accuracy\n",
      "WARNING:tensorflow:Reduce LR on plateau conditioned on metric `val_acc` which is not available. Available metrics are: loss,accuracy,val_loss,val_accuracy,lr\n",
      "295/295 [==============================] - 74s 252ms/step - loss: 0.2001 - accuracy: 0.9368 - val_loss: 1.2350 - val_accuracy: 0.6101\n",
      "Epoch 2/3\n",
      "295/295 [==============================] - ETA: 0s - loss: 0.0749 - accuracy: 0.9767WARNING:tensorflow:Early stopping conditioned on metric `val_acc` which is not available. Available metrics are: loss,accuracy,val_loss,val_accuracy\n",
      "WARNING:tensorflow:Reduce LR on plateau conditioned on metric `val_acc` which is not available. Available metrics are: loss,accuracy,val_loss,val_accuracy,lr\n",
      "295/295 [==============================] - 74s 251ms/step - loss: 0.0749 - accuracy: 0.9767 - val_loss: 0.1058 - val_accuracy: 0.9707\n",
      "Epoch 3/3\n",
      "295/295 [==============================] - ETA: 0s - loss: 0.0581 - accuracy: 0.9817WARNING:tensorflow:Early stopping conditioned on metric `val_acc` which is not available. Available metrics are: loss,accuracy,val_loss,val_accuracy\n",
      "WARNING:tensorflow:Reduce LR on plateau conditioned on metric `val_acc` which is not available. Available metrics are: loss,accuracy,val_loss,val_accuracy,lr\n",
      "295/295 [==============================] - 74s 249ms/step - loss: 0.0581 - accuracy: 0.9817 - val_loss: 0.0611 - val_accuracy: 0.9817\n"
     ]
    }
   ],
   "source": [
    "# Fit the model\n",
    "history = model.fit_generator(train_gen, \n",
    "                              epochs = epochs, \n",
    "                              steps_per_epoch = train_steps,\n",
    "                              validation_data = test_gen,\n",
    "                              validation_steps = valid_steps, \n",
    "                              callbacks=[es, rp])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "6RtRMTmbt7s4"
   },
   "source": [
    "### Plot CNN model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 592
    },
    "id": "zu5qWyG6sH84",
    "outputId": "68b1bb9c-6443-46c6-d429-6c1c95117909"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1296x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Loss and accuracy curves for training and validation are plotted below\n",
    "fig, ax = plt.subplots(2,1, figsize=(18, 10))\n",
    "ax[0].plot(history.history['loss'], color='b', label=\"Training loss\")\n",
    "ax[0].plot(history.history['val_loss'], color='r', label=\"validation loss\",axes =ax[0])\n",
    "legend = ax[0].legend(loc='best', shadow=True)\n",
    "\n",
    "ax[1].plot(history.history['accuracy'], color='b', label=\"Training accuracy\")\n",
    "ax[1].plot(history.history['val_accuracy'], color='r',label=\"Validation accuracy\")\n",
    "legend = ax[1].legend(loc='best', shadow=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vq0aLo69uCvV"
   },
   "source": [
    "### Confusion matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "c1xSh3tnuGCJ"
   },
   "source": [
    "Confusion matrix brings model shortcomings to light. The confusion matrix plotted below is going to show us the validation results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 606
    },
    "id": "rOH8nMUesH_T",
    "outputId": "4c76d299-f546-439c-da19-e101159eaabf"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkwAAAJNCAYAAADDOCpmAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAABFQ0lEQVR4nO3deXhU5f3+8fuTDQQEEUlAiAgGq2xVCQgqqyIIyOpat6KYigtuWEWUVr4/adVuWusCqHVfqqIIFG0RBFfADXAHFwiSBARE2ZJMnt8fGVPUwBPITM7Myft1XV5mJjM594dnZnLnnJOJOecEAACAXUsJOgAAAECiozABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIBHWtABdmWfk/4ayvc72PjClUFHAAAAu1A3TVbZ9exhAgAA8KAwAQAAeFCYAAAAPChMAAAAHhQmAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAI9aU5hSUkxv3HmWnvn9UEnS5At66L0p52nRXWfryRtPVqP6dSpuO+60Llp+3yi9P/U8nXBUq6AiV8vEG8ard4/uGjF0cNBRYu61hQs0ZFB/DR7QT/dNnRJ0nJhhruQT1tmYK/mEdbZE+l5WawrTpUOP1CerNlRcnvvuV+p80UPqevEj+mzNRl1zehdJ0mEH7a9Te/1CR130kIbcMF23X9pXKSkWVOy9NnTYCN1977SgY8RcJBLR5Jsn6a57pmn6jFmaM3umVq5YEXSsamOu5BPW2Zgr+YR5tkT6XlYrClOLAxpoQNfWeuDF5RXXzX1nlSJlTpK06OO1anFAA0nS4G6H6F+vfKLikoi+KtyslV9vUpdDmwWSuzo653ZRw0aNgo4Rc8uXLVV2diu1zM5WekaGBgwcpPnz5gYdq9qYK/mEdTbmSj5hni2RvpfFrTCZ2WFmdq2Z3WFmt0c/Pjxe29ud237TWxPuW6iyaEH6qXNP7KAXF38pSWrRpIHy131X8bk167/XgdEyheAVFRaqWfP/FdjMrCwVFhYGmCg2mCv5hHU25ko+YZ4tkcSlMJnZtZKekGSSFklaHP34cTO7Lh7b3JWTurZW0aatendFUaWf/+0ZXRWJlOmJeR+XX1HJ0TfnKi9aqHlOP18Ls+Q7ZPpTzJV8wjobcyWfMM+WSNLi9HUvkNTeOVey85Vm9hdJH0j6Y2V3MrM8SXmSlNb+VKVld692kO7tDtTgbm00oMvBqpOepob1MnT/NQN0/m1zdNYJ7TSwa2udNP6ZituvWf+9Wjbdt+JyiwMaaO03W6qdA7GRldVMBWsLKi4XFRYqMzMzwESxwVzJJ6yzMVfyCfNsiSReh+TKJB1YyfXNo5+rlHNuinMu1zmXG4uyJEkT//macs6ZpsN+fb/O/eNszX9/tc6/bY76dW6lq0/N1Sk3zdC2HaUVt5/15uc6tdcvlJGeqlZZDZVzYGMt/rRgN1tATWrfoaNWrfpS+fmrVVJcrDmzZ6lXn75Bx6o25ko+YZ2NuZJPmGdLJPHaw3SFpLlm9pmk1dHrDpKUI+nSOG1zj/z14r6qk56qmTePkCQt+rhAY++cq49WfaNnFn6qd+89V6WRMl1x18u7PPcpkV077iotWbxImzZtVL++PTXmkss0YuSpQceqtrS0NI2fMFFj8karrCyiYcNHKienbdCxqo25kk9YZ2Ou5BPm2RLpe5nF6/wcM0uR1FVSC5WfGZQvabFzLlKV++9z0l+Tr6VUwcYXrgw6AgAA2IW6aZWdzRy/PUxyzpVJejNeXx8AAKCm1Ir3YQIAAKgOChMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeFCYAAAAPChMAAAAHhQmAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4mHMu6AyV2l6qxAxWTY27jg06QlxsXHRH0BEAAKi2ummyyq5nDxMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeFCYAAAAPChMAAAAHhQmAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8an1hem3hAg0Z1F+DB/TTfVOnBB1nr6SkmN547Ld65va8iuvGnN5T7z87QW//a7xuvnyIJGn/RvU0597LtO7V2/TXa08JKm61hWHNKsNcySesszFX8gnrbIk0V60uTJFIRJNvnqS77pmm6TNmac7smVq5YkXQsfbYpWf21idfFFRc7pnbVoN7d1SX029R51P/oL899LIkafuOUk26e5bG//W5gJJWX1jW7KeYK/mEdTbmSj5hnS3R5qrVhWn5sqXKzm6lltnZSs/I0ICBgzR/3tygY+2RFpn7aUCPdnrguTcqrss75Tj96YH/qLikVJK0buP3kqSt24v1+nufa3txSSBZYyEMa1YZ5ko+YZ2NuZJPWGdLtLlqvDCZ2aia3uauFBUWqlnzZhWXM7OyVFhYGGCiPXfbuBGacPsMlZW5iutyWjXVsUcdogUPXqWXpo5V53YHBZgwtsKwZpVhruQT1tmYK/mEdbZEmyuIPUw3BbDNSjm5n11nZgEk2Tsn9Wivog3f6d2PVv/o+rTUFDXet556nvcXXf+35/TILQnTUast2ddsV5gr+YR1NuZKPmGdLdHmSovHFzWzpbv6lKSs3dwvT1KeJN1517264MK8Xd00JrKymqlg7f/O/SkqLFRmZmZctxlL3X/ZRoN7ddSA49qpTka6Gtavq/v/3zlaU/Stnnv5fUnSkg9WqazM6YD9Gmj9pu8DTlx9yb5mu8JcySesszFX8gnrbIk2V7z2MGVJOlfSyZX8982u7uScm+Kcy3XO5ca7LElS+w4dtWrVl8rPX62S4mLNmT1Lvfr0jft2Y2XinS8o56SJOmzwTTp3/D81f8mnOv+Gh/XCvKXq3eVQSVLOQU2VkZ4airIkJf+a7QpzJZ+wzsZcySessyXaXHHZwyRppqQGzrn3fvoJM5sfp23usbS0NI2fMFFj8karrCyiYcNHKienbdCxqu3B59/Uvb//lZY8dZ2KSyIa/btHKj738czfad/6dZWRnqaTe3fS4Ivv0sc7/YZdogvrmjFX8gnrbMyVfMI6W6LNZc79/BhhItheWsnByxBo3HVs0BHiYuOiO4KOAABAtdVNU6UnStXqtxUAAACoCgoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeJhzLugMldpeqsQMVk1lCfrvXV1NTp0adIS42fh0XtARgFAL6cuiJMks6ATxEeY12yddla4ae5gAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeFCYAAAAPChMAAAAHhQmAAAADwoTAACAB4UJAADAg8IEAADgUasL08Qbxqt3j+4aMXRw0FHiIhKJ6IxThmvsxb8JOspeSUkxvfGXEXpmQn9JUuMGdTTz9wO17K7TNfP3A7Vf/QxJ0hk9c/TmX0dU/Lfl2QvVqXWTIKPvldcWLtCQQf01eEA/3Td1StBxYiasc0nhnS2Mc+3YsUNnnXGKThsxRCOGDtJdd94RdKSYCuOaFaxdq9GjztHwk0/SiKGD9OjDDwaap1YXpqHDRujue6cFHSNuHnvkIbVu0yboGHvt0sEd9En+porL40YeoflL16jjxU9q/tI1GjfyCEnSEwtWqNuVz6rblc/qgr/N01dF32npF98EE3ovRSIRTb55ku66Z5qmz5ilObNnauWKFUHHqrawziWFd7awzpWRkaGp9z+op56doSeffk6vv7ZQS99/L+hYMRHWNUtNS9XV11yn6S/8Ww8/9qSefOIxrVwZ3FxxK0xmdpiZHW9mDX5y/YB4bXNPdc7tooaNGgUdIy4KCwr06oJXNHzkqUFH2SstmtTXgNyD9MB/Pq64bnDXVnpk3qeSpEfmfaqTjz74Z/c7rUeOnlq4sqZixszyZUuVnd1KLbOzlZ6RoQEDB2n+vLlBx6q2sM4lhXe2sM5lZqpXr74kqbS0VKWlpTKzgFPFRljXrGnTTB3err0kqX79BmrTpo2KCgsDyxOXwmRmYyU9L+kyScvNbOhOn54cj23ix267ZbIuv2qcUpL0BeG2C7prwoNvqcy5iusy99tHBRu3SZIKNm5T00b7/Ox+pxx3iJ5amHw/WRUVFqpZ82YVlzOzslQY4AtDrIR1Lim8s4V1Lql8T8xpI4eqb89j1K37MerY6ZdBR4qJMK/ZD9asydfHH30U6JrFaw/ThZI6O+eGSeot6UYzuzz6ueT8Dp5EFsyfp/33b6J27TsEHWWvnJR7kIq+3aZ3V67fo/t1adtUW3eU6sNVG+OULH6c3M+uC8NPv2GdSwrvbGGdS5JSU1P11DPP68W5r2j5sqVa8dmnQUeKiTCvmSRt3bpF464cq2uuvV4NGjTw3yFO0uL0dVOdc99LknPuSzPrLelpM2ul3RQmM8uTlCdJd951ry64MC9O8cLtvXff0SvzX9arC19R8Y5ibdnyvSZce41uvuW2oKNVSffDsjS4SysN6HyQ6qSnqmG9DN1/RR8VbdqmZo3L9zI1a7yP1n277Uf3O7VHTlLuXZKkrKxmKlhbUHG5qLBQmZmZASaKjbDOJYV3trDOtbOGDRsqt8vReu3Vhcppe2jQcaotzGtWUlKiq68Yq4GDTtbx/U4MNEu89jAVmNkRP1yIlqfBkg6Q1HFXd3LOTXHO5TrncilLe2/slVfrxbmvaPZLL+uPt/1ZXboenTRlSZImPrJYOaMf02F5j+vcP8/V/KVrdP7f5mnWoq90dp/yF7ez+xyqmYu+qriPmTTimNb6VxKevyRJ7Tt01KpVXyo/f7VKios1Z/Ys9erTN+hY1RbWuaTwzhbWuTZs2KDNmzdLkrZv36633nxdrVsn7y/F7Cysa+ac000TJ6h1mzY657xRQceJ2x6mcyWV7nyFc65U0rlmdm+ctrnHrh13lZYsXqRNmzaqX9+eGnPJZRqRpCdJ1wZ/evY9PXLNCTrvhMO0ev33OuvW/1Z87rj2zbXmmy36svC7ABPuvbS0NI2fMFFj8karrCyiYcNHKienbdCxqi2sc0nhnS2sc61fV6QbJ1ynskhEZc7pxP4D1LN3n6BjxURY1+y9d9/WzBeeV9u2h+q0keWnQl92+VXq0bNXIHnMuZ8f+0wE20srOSgbAmUJ+u9dXU1OnRp0hLjZ+DR7O4F4CunLoqTyvd9hFOY12ye98lOHavX7MAEAAFQFhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeFCYAAAAPChMAAAAHhQmAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAe5pwLOkOltpcqMYOh1mk86E9BR4iLjbPGBR0hbhL0Za3azIJOAIRf3TRV+kxjDxMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeFCYAAAAPChMAAAAHhQmAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB41PrC9NrCBRoyqL8GD+in+6ZOCTpOzEy8Ybx69+iuEUMHBx0lpsIwV0qK6Y1/nKNnJg2XJE0e3UvvTRulRXefpycnDlWj+nV+dPvspvtq3XNjdcUpuUHErbawPsd27Nihs844RaeNGKIRQwfprjvvCDpSzIR1zcLw+rErrFn81erCFIlENPnmSbrrnmmaPmOW5syeqZUrVgQdKyaGDhuhu++dFnSMmAvDXJcOO0qfrN5QcXnuO1+qc94/1XXMg/pszUZdc8bRP7r9rRf10UuLv6jpmDER5udYRkaGpt7/oJ56doaefPo5vf7aQi19/72gY1VbmNcsDK8flWHNakatLkzLly1VdnYrtczOVnpGhgYMHKT58+YGHSsmOud2UcNGjYKOEXPJPleLAxpoQNc2euDfSyuum/vOV4qUOUnSoo++VosDGlR87uTuOfpi7bf68KtvajxrLIT5OWZmqlevviSptLRUpaWlMrOAU1VfmNcs2V8/doU1qxlxK0xm1tXMukQ/bmdmV5nZwHhtb28UFRaqWfNmFZczs7JUWFgYYCKE3W0X9dWEaQsU7Uc/c27/jnoxujepXp10XX1aV938yOs1mDC2wv4ci0QiOm3kUPXteYy6dT9GHTv9MuhI1Rb2NQsj1qxmxKUwmdnvJN0h6W4z+4OkOyU1kHSdmU2Ixzb3htPPv2uF4SdEJKaTjm6jok1b9e6Kyl/Ifnvm0YpEyvTEyx9Jkm489xj9ffrb2rK9pCZjxlTYn2Opqal66pnn9eLcV7R82VKt+OzToCNVW9jXLIxYs5qRFqeve4qkIyTVkVQgqaVzbrOZ3SbpLUk3V3YnM8uTlCdJd951ry64MC9O8cplZTVTwdqCistFhYXKzMyM6zZRe3Vv10KDux2iAV1aq05GmhrWy9D9vx2o82+drbNOaK+BXQ/RSdc9VXH7Loc11/DjDtXNF/RUowZ1VOacthdHdM+MdwOcYs/UludYw4YNldvlaL326kLltD006DjVUlvWLExYs5oRr8JU6pyLSNpqZiudc5slyTm3zczKdnUn59wUSVMkaXtpJZU5xtp36KhVq75Ufv5qZWVmac7sWfrDbX+O92ZRS018YKEmPrBQktSjU7auOCVX5986W/1yD9bVp3XVidc8oW07Situf8LVT1R8POHsY7Rle3FSlSUp3M+xDRs2KC0tTQ0bNtT27dv11puva9T5FwYdq9rCvGZhxZrVjHgVpmIzq+ec2yqp8w9XmlkjSbssTDUtLS1N4ydM1Ji80Sori2jY8JHKyWkbdKyYuHbcVVqyeJE2bdqofn17aswll2nEyFODjlVtYZzrr5ccrzrpqZr5h/I5Fn38tcbe8d+AU8VGmJ9j69cV6cYJ16ksElGZczqx/wD17N0n6FjVFuY1C+Prh8Sa1RRzLvY7csysjnNuRyXXHyCpuXNume9r1MQeJqAqGg/6U9AR4mLjrHFBR4ibOLysJQROSwHir26aKn2mxWUPU2VlKXr9eknr47FNAACAeKnV78MEAABQFRQmAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8DDnXNAZKrW9VIkZDAiJxsPuDDpC3Gx87tKgIwBIUnXTZJVdzx4mAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeNT6wvTawgUaMqi/Bg/op/umTgk6TsxMvGG8evforhFDBwcdJebCumZhmCslxfTG7afrmYnlj7sRxx6it/9xprbMuERH5WRW3O6gzH214ZmL9OYdp+vNO07XHZf0Dihx9YRhzX6K147kFNbZEunxWKsLUyQS0eSbJ+mue6Zp+oxZmjN7plauWBF0rJgYOmyE7r53WtAxYi6saxaWuS4d8kt9snpjxeUPvtqgMyb/W69+8PXPbvt5wbfqNvZJdRv7pMb+Y34NpoyNsKzZT/HakXzCPFsiPR5rrDCZ2UM1ta2qWr5sqbKzW6lldrbSMzI0YOAgzZ83N+hYMdE5t4saNmoUdIyYC+uahWGuFk3qa0CXVnrgpQ8qrvskf6M+W7MpuFBxFIY1qwyvHcknzLMl0uMxLoXJzGb85L8XJI344XI8trk3igoL1ax5s4rLmVlZKiwsDDARfMK6ZmGY67a8Hppw/+sqc1W7/cFZDfXG7afrpT8M17Htm8c3XByEYc1qkzCvV5hnSyRpcfq6LSV9KGmaJCfJJOVK+nOctrdXnH7+ym5mASRBVYV1zZJ9rpO6HKyiTdv07sp16tGxhff2BRu26NBRD2rDd9t15CFN9dQNA3XUxY/pu20lNZA2NpJ9zWqbMK9XmGdLJN49TGZ2q5k1NLN0M5trZuvN7GzP3XIlvS1pgqRvnXPzJW1zzr3inHtlN9vKM7MlZrakJk5ay8pqpoK1BRWXiwoLlZmZuZt7IGhhXbNkn6t7u+YafHRrfXzfuXrotyeqd6cWuv/qfru8fXFpmTZ8t12S9O7Kdfq8YLPatmhcU3FjItnXrLYJ83qFebZEUpVDcic65zZLGiwpX9Khkq7Z3R2cc2XOub9KGiVpgpndqSrszXLOTXHO5Trnci+4MK8K0aqnfYeOWrXqS+Xnr1ZJcbHmzJ6lXn36xn272HthXbNkn2vig28o59f/1GEXPKRzb31J85eu0fl//s8ub39Aw7pKSSn/CfjgrIbKObCRvij4tqbixkSyr1ltE+b1CvNsiaQqh+TSo/8fKOlx59yGqu7qc87lSzrVzAZJ2rx3EeMnLS1N4ydM1Ji80Sori2jY8JHKyWkbdKyYuHbcVVqyeJE2bdqofn17aswll2nEyFODjlVtYV2zsM41pHsb/eU3PXVAo3307O8Ga+kX6zVk4gwd16GFbjyrq0rLnCKRMl32j/na+P2OoOPukbCuGa8dySfMsyXS49Gc2/0Zmmb2R0nDJG2T1FXSfpJmOueOjmew7aWVHJQFEDONh90ZdIS42fjcpUFHAJCk6qap0r1C3kNyzrnrJHWXlOucK5G0VdLQ2MYDAABIXFU56buepEsk3R296kCVn9QNAABQK1TlpO8HJBVLOiZ6OV/S/4tbIgAAgARTlcJ0iHPuVkklkuSc2yZVfnwPAAAgjKpSmIrNbB+VvwGlzOwQScn16ywAAADVUJW3FfidpDmSss3sUUnHSvp1PEMBAAAkkqq8meR/zOwdSd1Ufijucufc+rgnAwAASBDewmRmPaMffhf9fzszk3NuQfxiAQAAJI6qHJLb+c+g1FX5m1e+LYn3XQcAALVCVQ7JnbzzZTPLlnRr3BIBAAAkmKr8ltxP5UvqEOsgAAAAiaoq5zD9Xar4u24pko6Q9H4cMwEAACSUqpzDtGSnj0slPe6cey1OeQAAABJOVc5herAmggAAACSqXRYmM1um/x2K+9GnJDnnXKe4pQIAAEggu9vDNLjGUgAAACSwXRYm59xXNRkEAAAgUXnfVsDMupnZYjP73syKzSxiZptrIhwAAEAiqMr7MN0p6UxJn0naR9JoSX+PZygAAIBEUpW3FZBzboWZpTrnIpIeMLPX45wLAAAgYVSlMG01swxJ75nZrZLWSqof31gAAACJY5eH5MwsN/rhOdHbXSppi6RsSSPjHw0AACAx7G4P01QzayDpcUlPOOc+lHRTzcQCAABIHOZcZe9NGf2k2S8knSHpdEnF+l95ivtbDmwvrfRNMwHAq+XoJ4KOEBf5084IOgIQenXTZJVdv9vfknPOfeKcu8k5107SeZL2k/SymfG35AAAQK1RlbcVkJmlSMqUlKXyE77XxTMUAABAItntb8mZWQ+VvwfTMEnLJT0h6Urn3LfxjwYAAJAYdvfHd1dLWqXyknSTc66wxlIBAAAkkN3tYTqOvycHAACwm3OYKEsAAADlqnTSNwAAQG1GYQIAAPDY3Unff5d2/eaRzrmxcUkEAACQYHZ30veSGksBAACQwHZZmJxzD9ZkEAAAgES12zeulCQzayrpWkntJNX94XrnXN845gIAAEgYVTnp+1FJH0lqLekmSV9KWhzHTAAAAAmlKoWpiXPuPkklzrlXnHPnS+oW51wAAAAJw3tITlJJ9P9rzWyQpK8ltYxfJAAAgMRSlcL0/8yskaSrJf1dUkNJV8Y1FQAAQALxFibn3Mzoh99K6hPfOAAAAInHew6TmR1qZnPNbHn0ciczuyH+0QAAABJDVU76nippvKLnMjnnlko6I56hAAAAEklVClM959yin1xXGo8wAAAAiagqhWm9mR2i6N+VM7NTJK2NayoAAIAEUpXCdImkeyUdZmZrJF0h6aJ4hqpJry1coCGD+mvwgH66b+qUoOPETFjnksI7G3MlnjrpKXppYj/Nn9Rfr958kq4d1kGSNG3MMZo3qb/mTeqvd/50suZN6i9JSks13Tn6aC34vwF6ffJJunzQ4UHG32vJvGa7E9a5Jt4wXr17dNeIoYODjhJzibRmVfktuc8lnWBm9VVesLZJOl3SV3HOFneRSESTb56ke6c+oKysLP3q9FPUu09fHZKTE3S0agnrXFJ4Z2OuxLSjpEzDb5mnLTtKlZZqmnX9CfrvsrUafffrFbeZdMYR2ry1/O3qhnY5SHXSU9TzxjnaJyNVr00eqGffWqXV67cENcIeS/Y125WwziVJQ4eN0Jm/OlsTxl8bdJSYSrQ12+UeJjNraGbjzexOM+snaauk8yStkHRaTQWMp+XLlio7u5VaZmcrPSNDAwYO0vx5c4OOVW1hnUsK72zMlbi27Cg/ZTM9NUXpqSbn3I8+P7TLQXr2rfKfH51zqlcnTakpprrpqSopjei7bSU/+5qJLAxrVpmwziVJnXO7qGGjRkHHiLlEW7PdHZJ7WNIvJC2TdKGklySdKmmYc27onmzEzI4zs6vM7MS9ThoHRYWFata8WcXlzKwsFRYWBpgoNsI6lxTe2ZgrcaWYad6k/vrojmGa/0Gh3vl8Q8Xnuh/aVOs2b9fnhd9LkmYsWa2tO0r1wd+G6r2/DNE//v2JNm0pDir6XgnDmlUmrHOFWaKt2e4OybVxznWUJDObJmm9pIOcc9/5vqiZLXLOdY1+fKHKz4OaLul3ZnaUc+6P1Y9efU7uZ9eZWQBJYiusc0nhnY25EleZc+oz8UU1rJeuhy47Toe1aKSP13wrSRrR7X97lyTpqNZNFClz6nDl89qvXoZmXn+8XvmwQF+tS55DcmFYs8qEda4wS7Q1290epor9yM65iKQvqlKWotJ3+jhPUj/n3E2STpR01q7uZGZ5ZrbEzJbUxMldWVnNVLC2oOJyUWGhMjMz477deAvrXFJ4Z2OuxLd5a4le+7hIx3cs/4k3NcU0qHO2pr+1quI2I7u30txlBSqNOK3/bofe+my9jjh4/6Ai75UwrdnOwjpXmCXamu2uMP3SzDZH//tOUqcfPjazzb6va2aNzayJJHPOrZMk59wW7eY9nJxzU5xzuc653AsuzNvjYfZU+w4dtWrVl8rPX62S4mLNmT1Lvfr0jft24y2sc0nhnY25ElOTfeuoYb3yn//qpqeqZ7tm+mxt+c+NvdpnacXazVq7cVvF7fO/2aIeh5e/oNfLSFXuIU302Vrfy2ViSfY125WwzhVmibZmuzwk55xLrcbXbSTpbUkmyZlZM+dcgZk1iF6XENLS0jR+wkSNyRutsrKIhg0fqZyctkHHqrawziWFdzbmSkxZjerqzgu7KTXFlGLS84tW66X3v5YkDT+61Y8Ox0nS/XNX6I7RXfXqzSfJJD3+6hf6MP/bAJLvvWRfs10J61ySdO24q7Rk8SJt2rRR/fr21JhLLtOIkacGHavaEm3N7Ke/8RHXjZnVk5TlnPvCd9vtpZUcvASAKmg5+omgI8RF/jT+KhUQb3XTKt+x430fplhyzm2V5C1LAAAAiaQq7/QNAABQq1GYAAAAPChMAAAAHhQmAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwMOcc0FnqNT2UiVmMCAkEvSpHxNmQSeIj8anTAk6QlxsfDov6AhAhbppqvQVhD1MAAAAHhQmAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8Kj1hem1hQs0ZFB/DR7QT/dNnRJ0nJiZeMN49e7RXSOGDg46SsyFdc3COFfB2rUaPeocDT/5JI0YOkiPPvxg0JFiKtnXLCXF9MZfRuiZCf0lSY0b1NHM3w/UsrtO18zfD9R+9TMkSWmppqlje2vx7afo3b+fqnEjjwgw9d5L9vXanbDOlkhz1erCFIlENPnmSbrrnmmaPmOW5syeqZUrVgQdKyaGDhuhu++dFnSMmAvrmoV1rtS0VF19zXWa/sK/9fBjT+rJJx7TypXJP5cUjjW7dHAHfZK/qeLyuJFHaP7SNep48ZOav3RNRTEaeWwb1UlPVZfLn9YxVz+r0f0P10GZDYIJvZfCsF67EtbZEm2uuBQmMzvazBpGP97HzG4ysxfM7BYzaxSPbe6N5cuWKju7lVpmZys9I0MDBg7S/Hlzg44VE51zu6hho4T5p46ZsK5ZWOdq2jRTh7drL0mqX7+B2rRpo6LCwoBTxUayr1mLJvU1IPcgPfCfjyuuG9y1lR6Z96kk6ZF5n+rkow+WJDkn1aubptQU0z510lRcEtF3W0uCiL3Xkn29diessyXaXPHaw3S/pK3Rj2+X1EjSLdHrHojTNvdYUWGhmjVvVnE5MytLhSF5MQ+rsK5ZWOfa2Zo1+fr4o4/UsdMvg44SE8m+Zrdd0F0THnxLZc5VXJe53z4q2LhNklSwcZuaNtpHkvTs659r6/ZSffHA2fp06q/0t+eXauP3OwLJvbeSfb12J6yzJdpcaXH6uinOudLox7nOuaOiH79qZu/FaZt7zMn97DozCyAJqiqsaxbWuX6wdesWjbtyrK659no1aJBch3J2JZnX7KTcg1T07Ta9u3K9enRo7r19l7aZipSVqc35j6hxgzr67+Qhevn9Nfqy8LsaSBsbybxePmGdLdHmitcepuVmNir68ftmlitJZnaopF3uxzWzPDNbYmZLauLkrqysZipYW1BxuaiwUJmZmXHfLvZeWNcsrHNJUklJia6+YqwGDjpZx/c7Meg4MZPMa9b9sCwN7tJKH085Uw9dfbx6d2qh+6/oo6JN29SscflepWaN99G6b8v3Np3WM0cvvZuv0ojTum+3642PCtU5p2mQI+yxZF4vn7DOlmhzxaswjZbUy8xWSmon6Q0z+1zS1OjnKuWcm+Kcy3XO5V5wYV6cov1P+w4dtWrVl8rPX62S4mLNmT1Lvfr0jft2sffCumZhncs5p5smTlDrNm10znmj/HdIIsm8ZhMfWayc0Y/psLzHde6f52r+0jU6/2/zNGvRVzq7z6GSpLP7HKqZi76SJOWv+169Ox4oSapXJ01df5H5o5PFk0Eyr5dPWGdLtLnickjOOfetpF+b2b6S2kS3k++cS6iDqmlpaRo/YaLG5I1WWVlEw4aPVE5O26BjxcS1467SksWLtGnTRvXr21NjLrlMI0aeGnSsagvrmoV1rvfefVszX3hebdseqtNGDpUkXXb5VerRs1fAyaovjGv2p2ff0yPXnKDzTjhMq9d/r7Nu/a8k6Z5/f6Apl/XW23ecIjPTw3M/0fKvNgScds+Ecb1+ENbZEm0uc+7nxwgTwfbSSg5eAoiZBH3qx0QITt+oVONTwvP+Ojvb+HT8jygAVVU3TZW+gtTq92ECAACoCgoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeJhzLugMldpeqsQMVk0J+s9dbWZBJ8CeCutjUeLxmGwad7k06Ahxs3HxnUFHwB6qm6ZKX0HYwwQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeFCYAAAAPChMAAAAHhQmAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAeFCYAAACPWl+YXlu4QEMG9dfgAf1039QpQceJiR07duisM07RaSOGaMTQQbrrzjuCjhQzE28Yr949umvE0MFBR4mpsM5VsHatRo86R8NPPkkjhg7Sow8/GHSkmArj64cUjrlSUkxvPH6tnrn9oorrxpzRS+9Pv1FvPz1BN18+VJKU276V3nziOr35xHV668nrNKRPp6AiV0sY1qwyiTRXWqBbD1gkEtHkmyfp3qkPKCsrS786/RT17tNXh+TkBB2tWjIyMjT1/gdVr159lZSUaNS5v9JxPXqq0y+PCDpatQ0dNkJn/upsTRh/bdBRYiqsc6Wmperqa67T4e3aa8uW73XmaSPV7Zhjdcghyf0ck8L7+hGWuS79VR998kWh9q1fV5LUM7etBvfuqC6n/UHFJaVq2riBJOmDlV/r2LNuVSRSpmYHNNRbT47XrAXLFYmUBRl/j4RlzX4q0eaKyx4mMxtrZtnx+NqxtHzZUmVnt1LL7GylZ2RowMBBmj9vbtCxqs3MVK9efUlSaWmpSktLZWYBp4qNzrld1LBRo6BjxFxY52raNFOHt2svSapfv4HatGmjosLCgFPFRlhfP8IwV4vM/TTguPZ6YPrrFdflndpDf3rgPyouKZUkrdv4vSRp2/aSinJUJyNdzrmaD1xNYVizyiTaXPE6JPd/kt4ys4VmdrGZNY3TdqqlqLBQzZo3q7icmZWlwpC8mEciEZ02cqj69jxG3bofo46dfhl0JNRya9bk6+OPPgrNYzGsrx9hmOu2a0Zqwu3Pqazsf+Unp1Wmjj3yEC14aJxemna5Orc7qOJzXTq00ttPT9CSf12vsTc/kVR7l6RwrFllEm2ueBWmzyW1VHlx6izpQzObY2bnmdm+cdrmHnP6+U8SYdkTk5qaqqeeeV4vzn1Fy5ct1YrPPg06EmqxrVu3aNyVY3XNtderQYMGQceJibC+fiT7XCf16KCiDd/p3Y9W/+j6tNQUNW5YTz3P/ZOu/+tzeuTW8ys+t3j5V+p8ys067uxbdc35J6pORnKdrZLsa7YriTZXvAqTc86VOedecs5dIOlASXdJGqDyMlUpM8szsyVmtqQmTu7KymqmgrUFFZeLCguVmZkZ9+3WpIYNGyq3y9F67dWFQUdBLVVSUqKrrxirgYNO1vH9Tgw6TsyE9fUj2efqfkQbDe7VUR/PukkP/XGUenc5VPf/v3O1pnCTnpv7viRpyQdfqazM6YDGPy7vn3xRqC3bitU+58Agou+1ZF+zXUm0ueJVmH5UAZ1zJc65Gc65MyUdtIv7yDk3xTmX65zLveDCvDhF+5/2HTpq1aovlZ+/WiXFxZoze5Z69ekb9+3G24YNG7R582ZJ0vbt2/XWm6+rdes2AadCbeSc000TJ6h1mzY657xRQceJqbC+fiT7XBP/PkM5A27UYYN+p3Ove0DzF3+q8294SC/MX6reXQ+VJOUclKmM9DSt3/i9Wh3YRKmp5d8KD2reWIcenKWvvv4myBH2WLKv2a4k2lzx2u94+q4+4ZzbFqdt7rG0tDSNnzBRY/JGq6wsomHDRyonp23Qsapt/boi3TjhOpVFIipzTif2H6CevfsEHSsmrh13lZYsXqRNmzaqX9+eGnPJZRox8tSgY1VbWOd67923NfOF59W27aE6bWT5r3FfdvlV6tGzV8DJqi+srx9hnevB597Qvb8/S0v+db2KSyIaPfFhSdIxR7bRuFEnqqQ0orIyp8snP6lvNm0JOO2eCeuaJdpclqi/EbC9tJKDlyGQoP/c1RaCw+W1TlgfixKPx2TTuMulQUeIm42L7ww6AvZQ3TRV+gpS69+4EgAAwIfCBAAA4EFhAgAA8KAwAQAAeFCYAAAAPChMAAAAHhQmAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA9zzgWdoVLbS5WYwQAgIAn6cl1tZkEniJ8W5z8edIS4WHP/mUFHiJu6aar0EckeJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeFCYAAAAPChMAAAAHhQmAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMCDwgQAAOBBYQIAAPBICzpAkCbeMF4LXpmv/fdvomefnxl0nJh6beEC3fLHm1UWKdPwkafqggvzgo4UM2Gcjcdi8gnrmu3YsUPnn3eWSoqLVRqJ6IR+/XXxpWODjhUTyfxYrJOeopnXn6CM9BSlpaRoxuJVumX6cnU4aD/9+dddVCc9VZGyMl3z4BK98/kGHdVmf/1lVFdJkpl06/TlmvV2fsBT7LlEep7V6j1MQ4eN0N33Tgs6RsxFIhFNvnmS7rpnmqbPmKU5s2dq5YoVQceKibDOxmMx+YR1zTIyMjT1/gf11LMz9OTTz+n11xZq6fvvBR2r2pL9sbijpEzD/viyet0wR71u/LeO79RcuYc00e9PP0K3PrdcvW+coz88s0y/O/0ISdJH+d/q+N+9qN43ztFpt83Xn0d1UWqKBTvEXkik51mtLkydc7uoYaNGQceIueXLlio7u5VaZmcrPSNDAwYO0vx5c4OOFRNhnY3HYvIJ65qZmerVqy9JKi0tVWlpqcyS7xvtT4XhsbhlR6kkKT01RWmpKXJOcpL23SddktSwXoYKNm2TJG0rjihS5iRJddJT5VwgkastkZ5ncTkkZ2YZks6Q9LVz7r9m9itJx0j6SNIU51xJPLaLckWFhWrWvFnF5cysLC1bujTARLET5tnCiPVKTpFIRGeeNkKrV63S6Wf+Sh07/TLoSNUWhsdiiplentRfrbMa6P7/fqa3P/9GEx59R/+6prcmnXGEUsw04P/+U3H7zm2a6I7RR6vlAfV08b1vVhQo7J147WF6QNIgSZeb2cOSTpX0lqQukhJj31qIOf38SRGGnxClcM8WRqxXckpNTdVTzzyvF+e+ouXLlmrFZ58GHanawvBYLHNOvW+co45XPK8j2zTRYS0aaVTfHN3w6DvqdOUMTXjsHd0x+uiK27/9+Tc69vrZ6vf7l3TF4Haqk16rDypVW7z+9To6506XNFzSiZJOcc49LGmUpCN3dSczyzOzJWa25L6pU+IULfyyspqpYG1BxeWiwkJlZmYGmCh2wjxbGLFeya1hw4bK7XK0Xnt1YdBRqi1Mj8XNW0v02sdFOr5Tc51xXGu9sKT8ZO7nF63WUW2a/Oz2n369WVt2lOrwlvvVcNJwiVdhSokelttXUj1JPxyArCMpfVd3cs5Ncc7lOudyk+m3FxJN+w4dtWrVl8rPX62S4mLNmT1Lvfr0DTpWTIR5tjBivZLPhg0btHnzZknS9u3b9dabr6t16zYBp6q+ZH8sNtm3jhrWK//2WTc9Vb3aZ+mztZtVsGmbjj2svPj1bJellQXfSZIOOqB+xUneLZvUU9vm+2rVuu+DCR8S8XpbgfskfSwpVdIESf8ys88ldZP0RJy2uceuHXeVlixepE2bNqpf354ac8llGjHy1KBjVVtaWprGT5ioMXmjVVYW0bDhI5WT0zboWDER1tl4LCafsK7Z+nVFunHCdSqLRFTmnE7sP0A9e/cJOla1JftjMWu/ffSPvG5KNVNKivTcW6v00ntf69utxZp8VmelpZp2lER01QOLJEndDm2qywe3U0mkTGXO6ZoHl2jD98UBT7HnEul5Zi5Op86b2YGS5Jz72sz2k3SCpFXOuUVVuf/20koOOANALZasv+nkk2SnEu2RFuc/HnSEuFhz/5lBR4ibummq9BEZtzeudM59vdPHmyQ9Ha9tAQAAxBOnzAMAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeFCYAAAAPChMAAAAHhQmAAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMDDnHNBZ6jU9lIlZjAgJEoj4X2KpaVa0BGwBxL021BMWEgfis3OeyToCHGz6dGzK1019jABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeFCYAAAAPChMAAAAHhQmAAAADwoTAACAB4UJAADAo9YXptcWLtCQQf01eEA/3Td1StBxYmbiDePVu0d3jRg6OOgoMRfWNQvTXDdNvF4n9DpGpw0/ueK6b7/dpIvzztewwf11cd752rz52wATxkaY1mxnYZyrYO1ajR51joaffJJGDB2kRx9+MOhIMZXMa1YnPUVzJw3Qq5MH6Y1bBmv8yE6SpI6tGus/N/XXwskDNe//TtJRbZpIktJSTXf/prte++MgvXXrybpySPsayVmrC1MkEtHkmyfprnumafqMWZoze6ZWrlgRdKyYGDpshO6+d1rQMWIurGsWtrlOHjJcf7976o+u++d9U9Xl6G56buaL6nJ0N/3zvqm7uHdyCNua/SCsc6Wmperqa67T9Bf+rYcfe1JPPvGYVq5M/rmk5F+zHSVlGnLzf3Xc9bPU4/pZOr7TgcrNOUA3nXmkbnl2mXpcP1uTn35fk848SpI07OhWykhP1bHXzVLvG2ZrVN+2OuiA+nHPGbfCZGaHmNk4M7vdzP5sZheZWaN4bW9vLF+2VNnZrdQyO1vpGRkaMHCQ5s+bG3SsmOic20UNGyXUP3dMhHXNwjbXUbld1Ognj79X5s3V4CHDJEmDhwzT/Jf/G0Cy2Anbmv0grHM1bZqpw9uV74moX7+B2rRpo6LCwoBTxUYY1mzLjlJJUnpqitJTU+Sck3PSvvukS5Ia1svQ2k1bJUnOSfXrpCk1xVQ3I1XFpWXavK0k7hnjUpjMbKykeyTVldRF0j6SsiW9YWa947HNvVFUWKhmzZtVXM7MylJhSJ5AYRXWNQvrXDv7ZsM3ato0U1L5N68NGzYEnKh6wrpmYZ1rZ2vW5Ovjjz5Sx06/DDpKTIRhzVLMtHDyQH129ymat3yt3l75jcY/vESTzjxKy+8Yrv/71VGa9OR7kqTnF32lLTtK9ck/Rmr57SP091kfatOW4rhnTIvT171Q0hHOuYiZ/UXSbOdcbzO7V9Lzko6M03b3iJP72XVmFkASVFVY1yysc4VZWNcsrHP9YOvWLRp35Vhdc+31atCgQdBxYiIMa1bmnHpcP1uN6qXrkSt76fCWjfTrvm014ZElmrF4tYYdfZD+fmE3DfvDXHU+5ABFypwOu/QZ7Vc/Q/++sb/mLy/QV+u+j2vGeJ7D9EMZqyNpX0lyzq2SlL6rO5hZnpktMbMlNXHSWlZWMxWsLai4XFRYqMzMzLhvF3svrGsW1rl21mT/Jlq3rkiStG5dkfbff/+AE1VPWNcsrHNJUklJia6+YqwGDjpZx/c7Meg4MROmNft2a4le/ahQx3c6UGf0aKMZi1dLkp57a5WOOqT8pO9TjjlYc5d+rdKI0/rNO/TWp0U6sk38X0/iVZimSVpsZlMkvSHpTkkys6aSdrkf3jk3xTmX65zLveDCvDhF+5/2HTpq1aovlZ+/WiXFxZoze5Z69ekb9+1i74V1zcI618569u6rmTOekyTNnPGcevU5PthA1RTWNQvrXM453TRxglq3aaNzzhsVdJyYSvY1a7JvHTWqV74vpW56qnq1b67P1m5WwcZtOu7wLElSz/bN9HnBd5Kk/PVb1LNd+SHIenVSldv2AH329ea45zTnfr4rLyZf2Ky9pMMlLXfOfbyn999eWsk+xjhYuOAV3frHySori2jY8JG68DdjamKzcXftuKu0ZPEibdq0Ufs3aaIxl1ymESNPDTpWTIR1zWp6rtJI/J5i1//2Ki1ZslibNm1Uk/2b6DcXX6befY/XdeOuVEHBWjVr1ly3/PlvatRov7hsPy21Zg5H8FiMjTh9G/qRd99ZolHnnqW2bQ+VpZTvK7js8qvUo2evuG63po6M1fSaNTvvkZh9rfbZ++nui45RaorJzPTcW1/p1unL1O3QpvrjublKS0nR9pKIrn5gkd7/coPq10nTP37TXb9o0Uhm0qOvfK6/z/owZnk2PXp2pasWt8JUXTVVmIDaKp6FKWg1VZgQGwn6bSgmkuxUoiqLZWFKNLsqTLX6fZgAAACqgsIEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAeFCYAAAAPChMAAIAHhQkAAMCDwgQAAOBBYQIAAPCgMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KEwAAgAeFCQAAwIPCBAAA4EFhAgAA8KAwAQAAeFCYAAAAPChMAAAAHhQmAAAAD3POBZ0hcGaW55ybEnSOeAjrbMyVfMI6W1jnksI7G3Mln0SYjT1M5fKCDhBHYZ2NuZJPWGcL61xSeGdjruQT+GwUJgAAAA8KEwAAgAeFqVwoj/lGhXU25ko+YZ0trHNJ4Z2NuZJP4LNx0jcAAIAHe5gAAAA8an1hMrMBZvaJma0ws+uCzhMrZna/mRWZ2fKgs8SSmWWb2Twz+8jMPjCzy4POFAtmVtfMFpnZ+9G5bgo6UyyZWaqZvWtmM4POEktm9qWZLTOz98xsSdB5YsXM9jOzp83s4+hzrXvQmWLBzH4RXasf/ttsZlcEnSsWzOzK6GvHcjN73MzqBp0pFszs8uhMHwS9VrX6kJyZpUr6VFI/SfmSFks60zn3YaDBYsDMekr6XtJDzrkOQeeJFTNrLqm5c+4dM9tX0tuShiX7mpmZSarvnPvezNIlvSrpcufcmwFHiwkzu0pSrqSGzrnBQeeJFTP7UlKuc2590FliycwelLTQOTfNzDIk1XPObQo4VkxFX//XSDraOfdV0Hmqw8xaqPw1o51zbpuZPSVptnPun8Emqx4z6yDpCUldJRVLmiNpjHPusyDy1PY9TF0lrXDOfe6cK1b5wgwNOFNMOOcWSNoQdI5Yc86tdc69E/34O0kfSWoRbKrqc+W+j15Mj/4Xip9mzKylpEGSpgWdBX5m1lBST0n3SZJzrjhsZSnqeEkrk70s7SRN0j5mliapnqSvA84TC4dLetM5t9U5VyrpFUnDgwpT2wtTC0mrd7qcrxB8860tzOxgSUdKeivgKDERPWz1nqQiSf9xzoViLkl/k/RbSWUB54gHJ+klM3vbzAJ/Y70YaSNpnaQHoodRp5lZ/aBDxcEZkh4POkQsOOfWSPqTpFWS1kr61jn3UrCpYmK5pJ5m1sTM6kkaKCk7qDC1vTBZJdeF4qf6sDOzBpKekXSFc25z0HliwTkXcc4dIamlpK7R3dFJzcwGSypyzr0ddJY4OdY5d5SkkyRdEj0UnuzSJB0l6W7n3JGStkgKzfmdkhQ9zDhE0r+CzhILZtZY5UdHWks6UFJ9Mzs72FTV55z7SNItkv6j8sNx70sqDSpPbS9M+fpxW22pcOzGDLXoOT7PSHrUOfds0HliLXr4Y76kAcEmiYljJQ2JnuvzhKS+ZvZIsJFixzn3dfT/RZKmq/wwf7LLl5S/0x7Op1VeoMLkJEnvOOcKgw4SIydI+sI5t845VyLpWUnHBJwpJpxz9znnjnLO9VT5aSaBnL8kUZgWS2prZq2jP3GcIWlGwJmwG9GTo++T9JFz7i9B54kVM2tqZvtFP95H5S+AHwcaKgacc+Odcy2dcwer/Pn1snMu6X/ylSQzqx/9xQNFD1mdqPJDCEnNOVcgabWZ/SJ61fGSkvqXKipxpkJyOC5qlaRuZlYv+hp5vMrP70x6ZpYZ/f9BkkYowHVLC2rDicA5V2pml0p6UVKqpPudcx8EHCsmzOxxSb0lHWBm+ZJ+55y7L9hUMXGspHMkLYue7yNJ1zvnZgcXKSaaS3ow+ps7KZKecs6F6lfwQyhL0vTy709Kk/SYc25OsJFi5jJJj0Z/kPxc0qiA88RM9FyYfpJ+E3SWWHHOvWVmT0t6R+WHrN5VArwzdow8Y2ZNJJVIusQ5tzGoILX6bQUAAACqorYfkgMAAPCiMAEAAHhQmAAAADwoTAAAAB4UJgAAAA8KE4AaYWaR6F+IX25m/4r+evfefq1/mtkp0Y+nmVm73dy2t5kds9Pli8zs3L3dNoDaicIEoKZsc84d4ZzroPK/PH7Rzp+MvgfVHnPOjXbO7e6NFXtrp3c9ds7d45x7aG+2BaD2ojABCMJCSTnRvT/zzOwxlb8ZaaqZ3WZmi81sqZn9Rip/h3czu9PMPjSzWZIyf/hCZjbfzHKjHw8ws3fM7H0zmxv9A80XSboyunerh5n93szGRW9/hJm9Gd3W9Ojf5Prha95iZovM7FMz6xG9vn30uvei92lbk/9oAIJTq9/pG0DNM7M0lf8trx/eFburpA7OuS/MLE/lf2m9i5nVkfSamb0k6UhJv5DUUeXvsP2hpPt/8nWbSpoqqWf0a+3vnNtgZvdI+t4596fo7Y7f6W4PSbrMOfeKmU2S9DtJV0Q/l+ac62pmA6PXn6Dy8nW7c+6Hd8Heq71iAJIPhQlATdlnpz9ns1DlfxPwGEmLnHNfRK8/UVKnH85PktRIUltJPSU97pyLSPrazF6u5Ot3k7Tgh6/lnNuwuzBm1kjSfs65V6JXPagf//X6H/6w89uSDo5+/IakCWbWUtKzzrnA/hAogJpFYQJQU7Y5547Y+Yro32HbsvNVKt/j8+JPbjdQku/vOFkVbrMndkT/H1H0tdI595iZvSVpkKQXzWy0c66y8gYgZDiHCUAieVHSGDNLlyQzO9TM6ktaIOmM6DlOzSX1qeS+b0jqZWato/fdP3r9d5L2/emNnXPfStr4w/lJKv+jzq/89HY7M7M2kj53zt0haYakTns6IIDkxB4mAIlkmsoPf71j5buf1kkaJmm6pL6Slkn6VJUUG+fcuug5UM+aWYqkIpX/VfoXJD1tZkMlXfaTu50n6Z7oWxx8LmmUJ9/pks42sxJJBZIm7cWMAJKQORfLPdgAAADhwyE5AAAADwoTAACAB4UJAADAg8IEAADgQWECAADwoDABAAB4UJgAAAA8KEwAAAAe/x9ajybnuOmfuwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(10, 10)) # Set Figure\n",
    "\n",
    "y_pred = model.predict(X_test) # Predict class probabilities\n",
    "\n",
    "Y_pred = np.argmax(y_pred, 1) \n",
    "Y_test = np.argmax(y_test, 1)\n",
    "\n",
    "conf_mat = confusion_matrix(Y_test, Y_pred) # Confusion matrix\n",
    "\n",
    "# Plot Confusion matrix\n",
    "sns.heatmap(conf_mat.T, square=True, annot=True, cbar=False, cmap=plt.cm.Blues, fmt='.0f')\n",
    "plt.xlabel('Predictions')\n",
    "plt.ylabel('Real Values');\n",
    "plt.show();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Ccb3PmkTuNQ2"
   },
   "source": [
    "Prediction has been validated below"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 820
    },
    "id": "4ogPGyliniB5",
    "outputId": "40cb7312-7d15-4ef4-cd8f-c7b336d865ef"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x1008 with 16 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred = model.predict(X_test)\n",
    "X_test__ = X_test.reshape(X_test.shape[0], 28, 28)\n",
    "\n",
    "fig, axis = plt.subplots(4, 4, figsize=(12, 14))\n",
    "for i, ax in enumerate(axis.flat):\n",
    "    ax.imshow(X_test__[i], cmap='binary')\n",
    "    ax.set(title = f\"Real Number is {y_test[i].argmax()}\\nPrediction is {y_pred[i].argmax()}\");"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluating on unknown test data (from Kaggle)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Considering the test.csv from kaggle website - https://www.kaggle.com/competitions/digit-recognizer/data to predict the label column. Once we have the predicted labels, the output is uploaded to Kaggle for an accuracy score. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>pixel0</th>\n",
       "      <th>pixel1</th>\n",
       "      <th>pixel2</th>\n",
       "      <th>pixel3</th>\n",
       "      <th>pixel4</th>\n",
       "      <th>pixel5</th>\n",
       "      <th>pixel6</th>\n",
       "      <th>pixel7</th>\n",
       "      <th>pixel8</th>\n",
       "      <th>pixel9</th>\n",
       "      <th>...</th>\n",
       "      <th>pixel774</th>\n",
       "      <th>pixel775</th>\n",
       "      <th>pixel776</th>\n",
       "      <th>pixel777</th>\n",
       "      <th>pixel778</th>\n",
       "      <th>pixel779</th>\n",
       "      <th>pixel780</th>\n",
       "      <th>pixel781</th>\n",
       "      <th>pixel782</th>\n",
       "      <th>pixel783</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 784 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  pixel8  \\\n",
       "0       0       0       0       0       0       0       0       0       0   \n",
       "1       0       0       0       0       0       0       0       0       0   \n",
       "2       0       0       0       0       0       0       0       0       0   \n",
       "3       0       0       0       0       0       0       0       0       0   \n",
       "4       0       0       0       0       0       0       0       0       0   \n",
       "\n",
       "   pixel9  ...  pixel774  pixel775  pixel776  pixel777  pixel778  pixel779  \\\n",
       "0       0  ...         0         0         0         0         0         0   \n",
       "1       0  ...         0         0         0         0         0         0   \n",
       "2       0  ...         0         0         0         0         0         0   \n",
       "3       0  ...         0         0         0         0         0         0   \n",
       "4       0  ...         0         0         0         0         0         0   \n",
       "\n",
       "   pixel780  pixel781  pixel782  pixel783  \n",
       "0         0         0         0         0  \n",
       "1         0         0         0         0  \n",
       "2         0         0         0         0  \n",
       "3         0         0         0         0  \n",
       "4         0         0         0         0  \n",
       "\n",
       "[5 rows x 784 columns]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "digit_test = pd.read_csv(\"test.csv\")\n",
    "digit_test.head()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "digit_test = digit_test.values\n",
    "digit_test = digit_test / 255.0\n",
    "digit_test = digit_test.reshape(-1,28,28,1)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# predict test data\n",
    "# pred = model.predict_classes(digit_test, verbose=1)\n",
    "predict_x=model.predict(digit_test)\n",
    "pred = np.argmax(predict_x,axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ImageId</th>\n",
       "      <th>Label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   ImageId  Label\n",
       "0        1      2\n",
       "1        2      0\n",
       "2        3      9\n",
       "3        4      0\n",
       "4        5      3"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "id = np.arange(1,28001,1)#creating a data frame with imageid and predicted label to submit it on kaggle.\n",
    "kaggle_df_cnn = pd.concat([pd.Series(id),pd.Series(pred)],axis=1)\n",
    "kaggle_df_cnn.columns = ['ImageId','Label']\n",
    "kaggle_df_cnn.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "kaggle_df_cnn.to_csv('mnist_cnn.csv',index=False)#exporting is as csv file, submitting it on kaggle and to recieve a score. "
   ]
  },
  {
   "attachments": {
    "image.png": {
     "image/png": "iVBORw0KGgoAAAANSUhEUgAABIoAAADYCAYAAACEGAgaAAAgAElEQVR4nOzdfVjUVf7/8WdijAqNimJSIppYLVAtYKlsbpDlTZKTGVhJtN5QplbelBFl5raR3Zj98qZN1L5LsKtSKmqKugamC2YJuQK7BWQIrSaFMSE6JPL7YwYYEAQVRen1uC4vnZnPnPP+nBm4rs/Lc87nisrKykpEREREREREROQ3r01LFyAiIiIiIiIiIpcGBUUiIiIiIiIiIgIoKBIRERERERERERsFRSIiIiIiIiIiAigoEhERERERERERGwVFIiIiIiIiIiICKCgSEREREREREREbBUUiIiIiIiIiIgIoKBIREREREREREZu2Ldn5qVOnOHbsOGUnTmCxlHPy5EkqKk61ZEkiIiIiIiIiIk3m4NCGtm3bYjA40qFdO5yc2tOmzeU7L+eKysrKyovdqaX8V34uMWM2l+Ls1IEO7dthaOfIlW2vxMHh8h1MEREREREREfltqag4xa8nf8Vyopyy4ycoPVaG0ehMp45GDI5XtnR5Z+2iB0U//nSUEnMpnTsZ6Wi8SsGQiIiIiIiIiLQaFRWnKDH/wtGfzXQ0OtO1S+eWLumsXLSgyFL+Kz8c+RGDoyNdu3RWQCQiIiIiIiIirVZFxSl+/OkolvJyru7W9bKZXXRRgqLjJywcOnyELp070bHjVRe6OxERERERERGRS0JJyS/8dPRn3Lp3o307Q0uX06gLPq3HUv4rhw4fwbWri0IiEREREREREflN6djxKly7unDo8BEs5b+2dDmNuuBB0Q9HfqRL505c5ex0obsSEREREREREbnkXOXsRJfOnfjhyI8tXUqjLmhQ9ONPRzE4OmomkYiIiIiIiIj8pnXseBUGR0d+/OloS5dyRhcsKLKU/0qJufSy291bRERERERERORC6NqlMyXm0kt6CdoFC4p+LjHTuZNRdzcTEREREREREQEcHNrQuZORn0vMLV1Kgy5IinPq1CnM5lI6GrXkTERERERERESkSkfjVZjNpZw6daqlS6nXBQmKjh07jrNTB80mEhERERERERGx4+DQBmenDhw7drylS6nXBUlyyk6coEP7dheiaRERERERERGRy1qH9u0oO3Gipcuo1wUJiiyWcgztHC9E0yIiIiIiIiIilzVDO0cslvKWLqNeFyQoOnnyJFe2vfJCNC0iIiIiIiIiclm7su2VnDx5sqXLqNcFCYoqKk5pfyIRERERERERkXo4OLShouI3tJm1iIiIiIiIiIhcfhQUiYiIiIiIiIgIoKBIRERERERERERsFBSJiIiIiIiIiAigoEhERERERERERGwUFImIiIiIiIiICKCgSEREREREREREbBQUiYiIiIiIiIgIoKBIRERERERERERsFBSJiIiIiIiIiAigoEhERERERERERGwUFImIiIiIiIiICKCgSEREREREREREbNq2dAHnpMKC2WzBYDRicDjzc0XfZZNzyIC7nxfuHeq2AcbOhprnysyYMWK0HWc5asZS9ZqDAaPR7tg6znisxYy5rPbxteoEKCskK70Ai5sn3r1ca71Wq+067VuOmuG0cag5L4vZjKWibrWG2udtr8JCQXY6BSUdcb/VC3f70zjLvuqeo+WoGYuhZnwbev60fgBzXgaZh6Cbty+enevUbCkk64sCSjq64+/Vo8H6qp1prM22Wgz2bVhO/7xEREREREREWqHLc0aRg5nNM/0JXppd/VRu3CPcMnE9BVVPFKUQbfLnttDniZ43kbt9/HlibWFNGxkLuMV/Ael2zaa/688t72ZUHcB8f38GhT7Cw396hLv+4MMtUzdQcFro0vixRRtmcssfh/Hwnx6p/hP776r3WsiNn8ItPsMYN28ez4UGcItpAell9bf98J8e4eF5Oyiye+2WmVsx13teRSTPs71nRIBdDXG1zrtaWQbRw3wInjHPOmb9TCzKqIqorH3Nz7A7vk5f66b5c9sIu/5MS0ivTrhstQ5bQpb9GObH8bD/6eNe3U9FIasifLjtkZeJfvN5Rvv788SGouq3W76cx92/MzHtzXlEPz6sdp+nfcaNjbX1HG4JtquxaD3T/GeyrqZLERERERERkVbr8pxRhCtjnpnOsuAFrBoVwxjDVha9ZWZqbAieDgBmkl6dSqzvIvYlBmJ0sAYKwaEziPVbTbhH03sa81oiUf2Aoxt4wv9tEiffy1Svczj21ul8sCIE17pvyl7OE7OLiEjcy9SbDLZgJIhpfxvKZ094nd52PSzrZ/DSkFTeucd42jgNi05kGFD0UQS3bRpSfw02RZuWENMtms/iQ3AHcldOYdonaYT7BlK35YYEzFjGigdcoSKN6L7hJO2fjJ993YVLWJ02gbm3W6fsZK1fXn9oVSUjjsjtYcTnRBLgAJZd8xi9Ygu594bhSRHrliynW3Qy8Q/2gIo8Vk2dQVJaGH6B9VRsG+upSZlEXI91rCcN4+E3B7BvzkCq5x3lLWDa0kC2PdHABy0iIiIiIiLSSl2eM4oAvCbw+mO5zFm8geT3XyPpvkim+tou9S3ppK13ZepD1pAIwNAvjIkDM0hJP8epIYar6EgRFkvjh9Z7rMXCkaNmzEfNmI/WvJD7xRZyR04k/CZb7Q49GPPeXjaO7VOrSUtJ1XvNmOvUEPHsZLKem0ni4XM7tSrG7j0wpC1ncXwauUUWPB9czMYXmx4SAViOFWE+aqYobQfJnUPwq30amEYOIXalbQaUJY3VS70Yds8ZGnTtgR9xzH93A1n5Fgy3R7JxRRie1opx9zCQunwJsbvyKDrZhzHvJRJVX0hEzViPud72hEMPxowPw/LRbrLsjguaPh3vJTOYn9GUD1tERERERESk9bh8gyIM+D3+PMPWzWB8vCevTwmsmRFytIgCPHF1qX08BqDepWMNW/W8iWCTibsHRbBu8BxCbz7HY9PmEuzvzy3+/rWWQ5kPZUN319phjMF42n5IsRFV7/Vn2oY6Ydetk3nvmRKem51Awa9nd361ur19Djs/noBx1zwevtOHG4OjSCps/H329r7/PA//6RHGzU2Au3zxrLM9ULf7Qgjf8SHr8sG8NY5VD4QR2uMMDXqE8ffkRQwrTGDa/T709g8nZn9VgGMg4MVkPn7cSNq8Rxj0Ox+Co7Y2sDywgbG+EjhWJxByC+H1hQOJmbWAdDMiIiIiIiIivxmXcVAEdB5C+FhgzHhM3e2ed/XE0ymF7P/aBQAVRRTlgrFD7eTihF2ocOLk6V0EPBRJ1OOBGI96MfPZENzPsKHxGY8NjGbPtzkc+DaHA99G4md72t0zENKyyLU71PxNGqn7i2ptYB2xuuq9OdalXXV4hr3N3Iq5zFmR13CBjSkzY7ghhKj3Etnz1V5WDEjjibkbqBVL/drAv20CZixjY2IiG7clM/VYFNMS6tTjPJDQx4pYtj6Bdf+3gzGD/eh4pposZiydBxIxP5Zte3P4bIaB6IeXVAdtFrMB71GRvLcxlf9+FUPArinMWV//rLF6x/pQAfQwUndrb0NgJCvu2MJzb27lyJnqExEREREREWlFLu+gqErbdrUfO/himtyH2FfnkVxoAUsRWXELWHQ0jNDBtvkk1/sxximOZQl5WCrA8k0Cy1YZGONbe62U+80DCbhnMi88Zmb+ErtNo+txxmNrLT2ruTuY6+AQTN/OI/pv2ZgrwJKfwvwpESzLpVZ4UWvpWd21Z2BdRvXnObA7pYmDdrqsv5m4ZWocubbmLQBOBlsdffALMRC7IoHcMqAsj1Ur4jCE+NmWgdVmKcqj4BCYy06v1XvkBFwXRDGnaDKhAxu+kxyAefvL3BK8gOSjQIXFWlPnqpqyiTH588RK62dIhbUvo6H+Nusd63e34vf4SLxPO9pAwIy3CfoupdayNBEREREREZHWrHUERfXwfuxD4kfmMe2PPvT+XQAPr+/B6xvnEFCVIRiHMPfv07G8OYwb+/blxmHzsDz5IVF31be/jW2Z2/bXWJTW2L41DRxba+mZ3V29Og/h9b9Ph/dDuaVvX24MjqIgZDXvjao9a8h+6dkt09ZT75yZHiHMnRN42uyYpvJ+dDFRDksI/l1fevf154ndA3nvmSG2pVpGhs1ZzUzLPO726Utvn2FEW6bz9xeG1FrKlTwrgN7X9eXG/o8Q6xLJe4/WsyG0x0jC78Ea0DRyy3nj0Bm8d0c60/z70ruvD3e/bSFq8QRbsONFxOJIWGzixr596e0/ldTbFzNzaAO7KnUewusfz6FbQu2x/uDBBta+dfBl5msT6g3CRERERERERFqjKyorKyubu9GcvHz69jmLW4u1tDILdDjXeKUVqrBgOWmggYk5tpk9BgyNhDzNzVIGhg4NvWjB0vbi1yQiIiIiIiJyLi7V7KRtSxdwSVBIVJtDI4GLg+GcZy2djwZDIgBDy9QkIiIiIiIi0pq02qVnIiIiIiIiIiJydhQUiYiIiIiIiIgIoKBIRERERERERERsFBSJiIiIiIiIiAigoEhERERERERERGwUFImIiIiIiIiICKCgSEREREREREREbBQUiYiIiIiIiIgIoKBIRERERERERERsFBSJiIiIiIiIiAigoEhERERERERERGwUFImIiIiIiIiICKCgSEREREREREREbBQUiYiIiIiIiIgIoKBIRERERERERERsFBSJiIiIiIiIiAigoEhERERERERERGwUFImIiIiIiIiICKCgSEREREREREREbBQUiYiIiIiIiIgIoKBIRERERERERERsFBSJiIiIiIiIiAgAbVu6gBb1azlF32eSmp1OZtFuMo/DEUsOuaeANi4EGLpAex/udO2Lt9dA/K91wXBlSxctIiIiIiIiInJh/CaDIvN3u1m3K5alRVkUNHTQqWJSjxfD8RxSi4GvAZwJcg3n0dtHENTL+aLV21IsZjM4GTE4tHQlIiIiIiIiInIxXFFZWVnZ3I3m5OXTt49Hczd73sy5n7Dw0yUsPVZ63m25O43i2TsnYvJsnYFR0doIBs1MwTA+lj0vDsTQ0gWJiIiIiIhcqiqKSF8bR+z6FHJLoKNnICMfCmNMP9eWruz8mLNJXL6M1Sl5lHTsQ1DIRCLu8cLY2GSCCjO5Kev58KME9v4P3G8OIfRPIwnqYzy9i7wU1v1fAqv/XQjX+BH6QBj3Bfap3Yc5g9jlKRTV15dXCDOH9rDr2/6z6Ihn4L2EPxSCX/davZIev5zkI/U16E3o9CG4N3KKzeFSzU5+G0HRr4dISpjNiz/k1P/FOmeOBFz9HG+E3IV7K1uSZt4SxaAn1uMeuZqNj3k18V1FJEVNZFGWN1OXRTPsMv+dKCIiIiIi0qjCDTxhmkHS0dNf8nxsNRsjfS/P/3gvTGD88CiSj9V+2jA4mm1/DcG9obCoopDEp01M22Su84KBoDeSWPFATahT8FEEd89KwYIBV68+kJ9N0bF6+ihMYOwfo0itr7/HVnMg0tf677IMok2hxOQBnfvg7VJEVp4Z6EPE6kSi+lV9EoWsGhtEZFp9DU7g428j8WtwYJrPJZed2LT+pWeHPuPVxFdYery8nhcd8ew0gkd7DcD/eh/cuzpjtP8JtpRi/vEQe79JJ/W7laz9ubhO0FRO6g+vMGp5Om+aZhHkdkHP5KIyDo1m37fRZ/2+ksPZZO3vRknFBShKRERERETkkmImaeHzJB0Fz0dj+PuLgbg6gKUwhehxEcQunUHMiGSm3tTSdZ6tQmKfjiL5mIGg6ETeC+mD4WQeq6aaiNwexbQ4Pz5+tE+97zRveZvnNpnBdzoffzAZPyOQn8D44CiSZ80g9tbVhHsA5q3Mn5WChUDmJsdYn6soJDYiiDnbo5i/ZSjv3GObgXQ4j1TA/clYNv7Ju3aHhppZSll/m0FMHriPj2Xj8wMxOoB51zyCw5cTM2M5ps8mY313EXlpQI/JxCdOwKd2g5w+7+m3pVXf9czy7VqeWj379JCojQum615n+7htbB83jfCgAXhfWyckAjA4Y7y2L0FBY3hh3Fq+HBfL/103AM86o1Z0/BP+tPo5Vn1bXxh1YRRtiiLYZCJyUxFFacuZFm4i2DSFRWnWKKtoV81z0RvysFS/cSuRJhPBUVspMmcQOzOcYJOJsTOXk3r49PaDl2bUPFlRRGr8XJ4wmWzvWUBittmu3YnM/wIgjfkTTQSblpPe1BOq0/YTs+PqrSdyUxHmjDjbuYUzbWkaRXahVPpS6/tjMqBg+zxbe1OI3pCNubHwqsJM1oYFtrbrnF9TxsCSwaJQE8GmuXX+N6GQVU+YCDZFEJvX1AEREREREZFLXx7pCRYgkImPW0MiAEOPQGa+NoepT47EaLG7pmjKNQdgzt7AfNu1WnD4DObXuZ6puT4qrL7uidxU1EAfy0nOq92H5ZsNxCzdQG5ZA6eVvZ5lGcBNkUQ92Me6b62hD2MiI/EG0pdvIaveN1rYm7YBCxDxjC0kAvAIIeoZLyCD1Sm2i6KyEswAgUMYXjWpxqEHw+8JtI5BWfVVLOYj1jVinu6eGDsba//pYNd7SSEAw4YNrF66Zrx9KMMACs0118XmIo4AePagb932Ol+W87+aVesNig59wrMb3iHxVO2n3V2msX3cWt41DcCz01m22cmDINPrbB+3jDdc6uxNdGo3szbMJvHQeVXddGVFZO3PpmDrazw8dgnJWdlk7d/K/LGPEL1gLoPCl5Cam0fW/q3EPG2yJroAFSUc2Z9NVn4Sr44OJTotl4L92aSuncfYu+2mS9razyq2Pa4oZNWkIMbOjiO1rAeengZyti5hWnAo8zMs9VXYdJYMFpmsbScVWqCiiOT4uYy920RM9unnO3r0PFJzC8nan0bivHAGzdxK9a+94myy9meT/rcZ3B2RwN6imjEYvTS7gQIAikh8Oojgp5eQmFEEWMhcu4RpwUFEbqkau0bGwOCFn3ceWfvjSEqz+0Wcl8KHW7LJOuqLf/2hu4iIiIiIXJZ60CcQIIVl76dQYHdpZOwXxszp0wnvV5WWNOGaAyjaMIVBwTNYtDYLC2DJ2sCip03cNimBgqqwqPr66G3GRywnaX82R8pq95GUaz00Z+s8xt8dSvSXVcXlETtlBtHzZhC9qf7NWYqyMygADIHeeNq/0MePYT2AwgwyD9f3TjPmqufrbM/S0akbAFkZedbrN1dv/PoAKVv5tNB2UEUhn25KAfrg51Wzl4ml7BcAPLtaSP9oOfMXLGD+0gTS69TgefO9GICkpLTqUMiStoUkwDDSr+ZcqkKq61w58WUCMQsWMH/BclZ92byb1VyuWmdQdGw3r65+o05I5MyYm2LZ/ugoPM93HpmxL2MeXcv2m/5Ye4OrU7t5avU7pB5r6I3NLzW9G3O/2su+vTlse9EXyCNmrYEVX+1lT2omn0UHAhYS1+6ovWwuLYOOkan8NzWVfTnJzA0Ejm1gdUoDPxhFaWzZboGB0Wzctph35q9mz7pohg0dCIV5WFyHMC9xGTNvBRjIzGWJbEyc0KR1nQUfvcb8bAuGUW+zZ08SGzemsid2Au7HsoleahcCAanpHYn6PJM9qXs5kDyHIMCyPoHkOmUnmb3ZmGkdg32xE3AHcpdsaXCGk2XXkurpkRs/T2VjYhL7kucQhJlVr8ZZ0/LGxgADASPCMACJW3dX112wez1ZgPeEoXg30L+IiIiIiFyOXLlv+nT8nCD3bxH88Xc+3BY8hTnxG8iqE2I06ZrDksaiqK2YCWRu8l62JSaybU8y8wYbsGyPIrrOvj+p23/hkY17OfBtDisecK3uwzDqbXbuTGRjYiJ7NkcT5JRHzIzlZFUA9CDgoSF433QvpgH1byxbkJsCQIB7jzqvuOLqCZBCXmHdd1lf9/Szvsc+rKGikM2brG1iLrE+7+DF1GVvE+6bTuTwAOvsqUHDiNznS/iiZUy12yq3oMC6mVDsk8MYPWseixYuYdG8KEbfbWL+lzXpnHHoy/w9MpATq8K57W4TwaZh3DY2jhODI/k4ekjNkrJDhdb9jlZN5e7QKKIXLmHRwnlEhgYRvCCD85wKcdlrfUHRqUMkrp7N0lohkQdPDvqQN+7yaMZNxBzxvOsVNg0aVTsMObWWp1Z/QsGpht7XzO4ZSoDt2+55s99pz7nfPoQAgIoTdd44FNNg2y8Fhx4EDRlo/XdDy7Mc2ln//ncCqzdkU3DUAn1CeO+9Ocy81+s8xrWQ1E0ZgBdRk++tnqppvD2SbXv3sm/OHbXXh94zkqCq32UegQy1lX2iTt1B94zE0zYF0TjwDutUw2MN/7jvTYnDAoQ/OQHvqqmLHmG899Ve9iWG4VlB08bAdyjhTsD6NPZaAIpI3Z4B+BIaqOlEIiIiIiKtjeGmyXz8ryRWRN5LQB8DRdlbiZ09g+AAH0a/l10dOjTpmiN9B7HHgLFjGWO3HGvM+DAAkrbvrvUf6YwZT7hXzRWTtQ9XIkbdgcFsxnzUjNkpkND7gMLdZB4CMOA9fjEbE9/GVDcHagbeDz6PyQkKVoRz2+gZRC+Yy/hBw/jQIfC0/zgvyEon8zszHLNYZ09ZLHA0j8y07JrZU4DniEXMi5xM1HtJ/DcnhwM5e9k4JxDDsWwWPR9HbtWBlkL2fpGF+RhYjgEVZixYMGftJjXP7nqw10jeeyOSqZExbPtPDge+zWHfxjkEOVnIWvg8sd80/7hcTlpdUFT0r3d45Wf7vYKcGdPvdZ7p53JB+jP2m8a7vx+AfQ5b9PM7vPmv4gbfc1G1pUkhTru2jRzlei9z59+LKxksetrEH/196H1TAKNnJzS8rrVJbJuI0Q2DU+1XDI2uDzU07eSqAp4z1FDwra1F59oNGozWdaoGB5o2Bg6+mB7rASSQ8oUFju4mJQXwHUnQpbeZvYiIiIiINAdjH4Iee5v4bXs58J9UtsVMxs/JQvqboUTvstDUa46i72379zh1rH2p07sPQVAzG6dKW/trnao+ilgU7s8t/lV/AngiHiCNvHqXi9VzOm7W6TxHjtXdP8mCtQAvenRu4M2dh/DOp7HMHNoHMjYQm5CFccqHfPyUrzXkutYVI2DJWMD4qXGku0wg/quq2VN7iX/MlfT4KYx/t2Zmj/H6QMY8Np3w23tYr80cjHg/GknUTUDebvYetp534tOhRG83ExCdzH9TE9m4MZX/JkcT8EsK0Q9PJbHq/Dv3IeiBCcwcOxB320AbvcJs+yjlkfbv3/YStNYVFB3bzdL03bWWWPl5vMErgy7s7cjcg15hqYd9ClBOYnrsRV2CdjG4j3qbPV9l8lliLO+8MgGTh5n0+CiCo2ovDzs7Bow9ACxwstlKPesaOlaF8L+e+cimjIF3YAjuWFi1KxtzWjJJQNBDQ2svUxQRERERkcufOY/UXWmkZhTWBDgGVzwHT+edFwYCFmJ3ZdPUaw5Dh6us/zhZZ0VI1VKpM6rqow8zV+9l397T/8z0bdppuXtYV6tU7ydUxZJLdhqAH6etSqtSZsbc1puIRUns+zaH/6au5p2xvli+tu57FHSLNwYgK2U5uUDAhLDqFTE4GAkIm0AAkLsipXrDbMtR6+woS62VJEYMXaD6WrJoN4nbLUAYE0PsivMIYeIY4FgKibabP1Fmm21VZ9FJ1T5KlpO/7cVnrSooKvhXbO0lZ4aJ/OU+72ZcblalnPRNq0ivDoIc8btvNi/ad3RqLUv/ld/sPbcUS2EGqRviiNlpxv2mgZjGRvLOX+cQAFjWp9dM9bMefRahTx/8BhuANFZvt7slWF4co6/rS++xCRQ00znUYikka39VpGjE+1brb8zYTSk1v+DLUphzU196XzeP1IqzGIObAgntAZb4D3lp0wZgSM0yPxERERERaUXyWB0eztjwuayrtWePhYJc65WMt5uRpl5zGL39rFubbN1Bqt3Kjaxd67EA7rd60/CVRVUfeSRnFdW6i1fBri3szcrjSNV1Wq3rodMZ/AYyxglYn8Bmu/MqWBtPLGAIGYh/1fXv0Tyy7O6qVrB2Irf4+3P3a3Z7FJVlEPN+CuBL4K3WMzC0s/69Nyu31iwpS242ewE6G2zX8kWsm+nPLf4m5u+xO7IwhS0pgFMf3F0BQ9Vt7dPJ/c6uwYo8cm038zYabC1umskt/v4Ev1l7HyXrRtoGPHv8tq/f2rZ0Ac3mVBaJ/7G/QZ8jTw4Yg3ezR2HlpCdO5rFvcyj67ifWPjoZPyegTV/CB4zi/R1rq2c0Jf/nE9LvmoxfK4jjDBVZzH96LulO6ymaM5nA7r+QFb+cVMDw6B22taZVCXYa0VFzKRw6koixvpx573ADARPmEPRRFMlzQxmbG4afi5n0uDjSMRD0QCDuQPNO/Cti1RNBRKZA0PxUVoxyxX3U80T8Xygx8VMZ/UsYQR5QsD2OxGPg+eRQ/B2aOgYAXgyb4MX8uRtI3ATcM4yghqZlioiIiIjI5ct4B+GP9SFxaQqRwwNYPWQkAT0sZCUkkHzYAk6BRAy17lXalGsOPEJ44bE4Ri9dzvjQI0T8aSCu6QlErywEp3uJetDrjOVU9zE3lLGHIpl4eztyP1rG/PXZWHzn8NnHvtR3PXT6eQ1hypxA1s1KIXK4ibQxgXT7MYXY9dmAL1GP2TaGrsgg+o+hxBwzMPXjTGb6gvuoyYTPiyB2RTi3ZdzLGC8LyZu2knsUPJ98vnrvJe9RMwhaMoPk+KmM/mUCj9zrC/9ez4crNmDBQND0ENs1liv3hYcxJyWOmMdCOTImEHcn6zVjKgaC5kwgwAAY7iD8yT4kLsxgTmg4eTNCGGosYUvCAmIzgD6TCR9svTp1vWcs4XNTiF0RwegfwwjyMGDOiCN2FxgGz2HiwOafbnI5aTVBkeWrbfyf/Wwip8mM8XNs5l7sQiIAyypG/Q3Whk3GzwgGvzE8++VaZlXNNDq1lk+/mohfs9fRAjzC+Gu8medmLiFmVgQxABjxG/s277w40Jb0Ghn2zGLG5M5g1a44FmUYCRrr2/idz3qE8N7HMNPHto4AACAASURBVGfGXFbFL7FOqXTyYswbbzO3vl9a581At+7WXxCetr/p4EvU32MxzpzK/PXLbVMcjQRMj+Gdyb7W82vSGFh5BobgN3cu6YDpnjsaCctEREREROTyZMDv2dV87P4yr769gfS1y213WjbgOngCr78YSVB326FNuebAgN+zHxLvNJMnFmxg0awN1qN8w3jn/81hWGP/AW3fx9IoUpda2/QeOYfXo8Ns22HUcz1UD/cHFrHR8jwPz9tA4ops6zu7BxL1/iLCq+7T42DE1cMA+X70qNoWuEMgcz+NxXXmVObv2kBMBrbru9eIGmV3I6Tu97JiW1cWzZ7JovVLiFxPdR8z34tm6u0114KGwDnsjHfluZlLqmuhsy/hb7xG1ANVy8wM+D21mo3uC3jptThiZ6cRaxtjv7HR/PnZELyrOq+qMWomi+w+C7+x0bz+fAjuDo2Mcyt3RWVlZWVzN5qTl0/fPhd35970j+5glN0apTG3fcIbf3Buxh7qhERVDGNqZhUB5n/N5qY9n9W87r6A/AeacpP4y0iZdS2nwWjb5LmBYywGI4aiBMYHRJHcwGERq3OI6le3bUMjm1g3D4sFDPV1YzFjLmv8/BodAxERERER+U2xmK376DR6ndCUa44KC2azBToYMZ7L5VEjfTR4PdRQHQ4GjPUVUmHBgqH+87CYMZc14frubM61zIz5ZAO11D2uKdeWtr5b4tquJbKTpmglM4pybLucV7mL4Tdd/JAIwHjT3YTv+cyWXAKHs8jC77TbAF7WOhgxdmj8GANAB08Cn5zc4Pl7150w1JS2m0mDvxQNTfjldBHrFBERERGRy4PBaGzaHrlNueZwOM//QG+kjyaFRE2pw8HQ8Dk35Tyb0oe9Dsamrdho6jXb+Y5zK9Q6gqIfvuFL+53j2/vg2WxrfZoeEgFg9MCnPXDc9vjXTHJ/AO+rm6uey4zRl/DpTdxaX0RERERERERaVCvYZhk4kk+S/eNu1zfTrcjPMiQCwIO+3ewf76bgSLMUIyIiIiIiIiJyQbWKoMhiKa/1eFgnlwaOPBvnEhJZuXaqvSdRSZ36REREREREREQuRa0iKDL/cqj2E1c0sL7QnE7spiwsjbZ47iERQLsrat/l7OAvpY32KCIiIiIiIiLS0lpFUNQk5nQWxU1n9tfT+FPimcKi8wuJREREREREREQuV60iKDJe5Vb7ico6MZAtJHrTAlBO6rcNhUXNExKdqKy91KznVc15BzYRERERERERkQujVQRFBkPtpV5JPxfXPsDBAG3sj7GGRY9/kmMXFjXfTKKin9NrPe5Ypz4RERERERERkUtRqwiK6ObBMPvHR76hwP6xkzdTH3mHZ2ttXVRO8jeTeXZrPpZmXW6WT06tu5wNwL1bQ8eKiIiIiIiIiFw6WkdQdPX19LvS7vHxTHLNdY5x8mbqo0tOC4sSsybyp6XNuCeROZ/M43aPr/TB8+qzbENEREREREREpAW0jqCIvvh3t3/8Tzbvr+dOYw2ERanHmm/javP+bcTaP9HdG++zb6ZJSr/ezOIZYUwIC2PaCx+Qfrhp78tcFsbK/efZeW48E4bOZUeJ9WF5aSnlFXavV5RTerzedzauaDOvz91McT0vnX/t2awMs47ZaX+WZTd7vRdMxWHSP3iBaWFhTAh7lsWbcymtaPxtzVNrKWlz7yHsg1xbLeWUltbel+u078NZaPAz/voDwuy+c+fk+LnXJSIiIiIi8lvQSoIi8L5uFK52j1dlbau9/KyKkzdTwxbUCYvsnNfdzQ6xOeszu8eOPHmdz7k01Ljs95k0OwO/aX/lvZi/8tJYR1aOf4EdTUkASg/zU8O3fWua3iZeWvw4d3QEOMzW56LYap+2ZbzP8Pczzq3tinJ+OlJe/2vnXbsX9y/+K/9v8V/5f7OHc1X74URWPR7rdW5NnqneC6KYHS+MZ6XDWF6KWcF7i5/E76soJjUl6GqWWp0Z+Pi7vHq/p/Vh0Wb+/Jx9+FTP9+FsNPQZe4bwavV37txkvn8fMef4tRQREREREfktcHj55Zdfbu5Gi4+W0MWlU3M3e0Ztr76K8i8+IbXS9sSv6TgZHiTAzeH0gw1u3HbDLVyZvYVU+9kF5xUSgSV9ORNy/ktZ1RNtHuB50wDcrji39s6keGcs+3yeYvwAFxyudKS9683493Omg1MPOrWHQ1sXkt6mP706W4+3f3xkdyz5Pe/n6n3xJCR9xSEHN66/xnpnttKMBNKOefDr5/GsScrmCo+bcavIZVtcPJ+mF2Ps05cu7YA2h9kX/zWG/q7kr/yA9Wn7yS8u4rtjbvi32UvM37ey75sCKgqLMdzoTbd2wOEMNq78iB3/yqW823W4d7bb5Pvw56z5YC27co9xTZdydn8Ot4+4kfZ1zrvB2stz2bZkJ6duvpEuVcsQczcTk3oFN13vgv23wKGdI47tHHE8eYBP/+XIkPtvpmM7RxyvtJ7/xsMe/M7NsXo87B8Xf5lI4sfbSMs9xjXX9eYqR6D0a7Z9Vs7vPA+y7R9JpOVCb+8etK+KYYty2bYunk93ZNcz1m4Ub17GxsNu+PdpagLyXz551YBpQTAeVzrg0K4j7r79ub69M53cnHEoyWDNusO4+7jhCGD/uPRrtn0Gt95WwrYV1vHu6tGbTu0ADrJj5bcYux5k8wdr2VfchV7XuVC6O564dTvIr/Dgd+622nO2s+uoB7/rkMWaFWvYlX2An4pyKe/mTemWOt+HPh2hopT85AQSPtlBzveOXHO9W834VJSSsyWeNUlfccjRgw75CRT2DOe2a+qcdvV3rjdXnekcqfNdc7sO946OHNq6kPitX/Pt/0o4VOTIDVXHVinJYM1nx3A/+Tlr/pFE5ml1HiZzXQKJ/9xFfqkb7h4dcWxTTs7GJWS26U8vF4BiMuOX8Z+rqn72DpO2bDMl13vTTXvai4iIiIiInZbITpqi1cwooo03pt/ZL/IqZ+HuVWSdauB4o1/tmUXnGRJxKofY3WtrLWEL+t0I/C7QCDu7dmf/5gTSv6+ZHeJyQ388XKz//ik7kcyfao6v+zhz2bvscBzEaNPvKV//DLPWWNetlR/YScxb77PfdTDB/sXETJrEtHd24jbExEA2M+kvm7Eu6vuJzDVZ/IQjXW7wwsPZGQ+v2/C/zhk69sT/hi7QzQt/f0+6OALfJzJr+nrKf29i9HAX0qMmEZNhqz37fcKeSoKBJoJ9f2HNgni+O8O511u7Y0+6lL7Pxi9rxiNz40JKO/bkbK7Pyw/sZPeB0nofl259gWfXg3/IOIK77+bZpxI4VHXg9wms2enCQNNwfIo/YOwbuygHOJzIrEff4rvuwxkd0h8+eoo/by2uGevZ89nrYhu3JuuCW++dfLImu2a5mXNPfHy7W8/1+EF27zxI9UjUffz9ZmJiD3K9KYQh3bOYN2khmaUAP5G5cj6L1hzD3zQY5x3P8OykF/hHSX+Ch3uS/84kYvfXGRdHN6738uAqZw98/G+jV8d6vg+UkrlwEvOy3RgSYsKn5O+Mm55om4FUzI4XQnk714MhpkFcs3chi1MbOu+q71wj51j8KX+evh4CQnjI5MbuZ55izUG46rrbuL4bdLnhNvxvcDv9e3H8ILuXvczind25I2Q4Xb5+jUkLM6xtVhxm4/Rn+KT89wSHDMctO5qnF2ZQjiNujodZvN02m6vkK9YsSyR250Hr48OfszYVupzNxysiIiIiItKCWk9QBLj/IZzH7M/IsowX12XR4Eolox9Tx7zOk8bzDIkoJ33dK/zFvqM2o3jsDx7n2mCjHAdFET/RmY2zQxl+Xxgvzk8ks4l7FAGU9w8lfLAnLj19uf/5Sbi9n0CmLXToOngs9/friVtACCOuLeWP4ePw6dkTn4njuP/L3DohjiNuvr54dHTBo19//DxdwMUTP6+ecK0XfgG+uLUvJz02Hp+oOdzfrycunsOZ8vJgdq/aRSnlpK1JZOAs62tuXiamTBl8DrU74nf3cLbu2mO9sD++i21bb+eOfs03jeO7rzO4/vbB9O3ujNvgKOJiQnCretE1hPCJ/fHo6cnAyeMI3prBNwDdTfxl3V+JGOyJS3cvgkf1Z1t2fnWbXe9/kojhtnGre54HM8g8WM9eW/Tk/nej8TvwAZPuvYewSS+wcvvBpu1RBFDixt2Ph+DTszseg58kcvjnxG61fXlKfAl+/E48enoR/MBgMj2GEzHcEzfP4Tz6YE8yc+p8ydp3x6efJ106euIf0B8Pl3q+Dwc3s/j7sbz55J14dO+Jz7g5RHb6gI3ZwIHNxP7yOK8+eScePT3xm/gkD57vj01+Fjv6DiLwhu4497yT6f9Ywf09wdmzPz7XgptXf/yqQrW6OpoIn+iLW3dP7p48CZ+UPeQD5akrWOw2iekPWl8b+GQUI/69kI0Hwbnf7Vy/M8N6XMZuSh8MwePzDIqB0n/v4afA/jXfExERERERkUtcqwqKcBrAY34Dau1VlJ4/i9k7DzX4FroM4JkJ5xMSQcHOV3gqP9/uGUdMfuEEnEebTeHSfxwvrVjH5tXvEvH7gywe/xTbmhgW+Xl51jxo70lfzxy+s02Hcqy1Ws8Fx7rrv85aMf8rLOXjaLuNo19O5LtffqGcYn465IlHD7vDXbvT61xq9x3MQ59/xt7jUP7lZ+wNvg+/8669hs/9z8AHoZgeepYFyxLJtN8PytGR6m2vHOwjiFJ+2hnP4hcmWc/7te212nSsZ2WkVTmZ8S/w9Mos6t1RyNmTu2e+SdymTayIegDHHc8waUlG/cfW5elFL7sZLh5e/Uk7WPUz4li7Jmens5qRVa8jh8nM+oCnqzcOn8TiL0v5pQz4qZhv+vakJiZzwe3a8+zv5hBeclhByOgwXvzLB2z891ls3W3/OXbsTJeScixAadFhbrrB024senL9zQc5dARw8WVAx8/Zfxi++SqXAUNM+LGH/SXl7N+Tz4hBPc/zhERERERERC6eti1dQHNz/cM0ZueG89TPVZfMpaz68jm6tX+HZ/qdPmvjfJm/fIenvvys1sbZrp2m8ewfmr8ve+WlpdDe2XpR7+iCx+Aneal4Egs+L+ZuU+N9/1RSClSlBeWUl1xFl2YMVU7Xk4i3/krwaSHAYaAc6twxrfwM8USDtTt4MXBIFBs/L8aQ9hVDgqOa8wTg2jt57h93wvHD5KcmMC8smoh1Ufid4S2l2+fzbNog3nz+XaY4O8KXCxm0qymdOeL3/Ca21/eS7Y5yzs7WMXLs6cv9c2Zy6J5dfPOkL41un15UjP08pfLSX3Brb4CG596dv8CZLJ/V//TnvwTK69wx7Xz32nbozh1z4rijopziA7v4xythLJi4jumDzi/yqnu3tPLyqlCtO/6DSon59y6KM/ozcFp3ugz+hb99uQfHfw9iyIvn1a2IiIiIiMhF1bpmFAG0ccMU+krtJWjks3DnI8z6Z34zXgqXk/vP2dyzcy3ptfofxbuhI3C/wCObHxvG07G5NU9UHCQt5TA+19lCIgdH8nNt+6SUfM6OlNrv35a4mUO2C9/yzxNY6XY7/ud8NylHDA7FlNrfttwBHIuP2ma4dMd/UDmfbK2pt/zLD1iQkEu57bWVCTWzYfK3bmbvGXo7U+19h4eQmfAyK7NM3HHTOZyKA3yXm29dxlVxkJSqvWcoJXNZNBsPAO274xE4nDtcD/NTfSvD7JSX/ETXG7xws4U6OXt2nkNRdZRsZ97ot0izG+/Sz3ex42ZP60wsB+D7XL6zvX4oZTtptd6/ma2f2wqvOMzWhD0EB5zjHd+q+isp5pfqIKXO98Hbl/tTttfUW3GYbfMXkn646rV4Nn5fVdun7Kg3HaunzwbOsTTjAxYkHAQHR1w87yQ40IVDxbbzdXCktKSRD60eLv0GYUlIsO3lBHyfyMqU2/G/0frQrf9g8pe8xVbf2+gLuPj255v33yUzcBB9z7o3ERERERGRltPqZhQB4DSAF0Jn8cPqN0is3sy6lFX7w8k5Mos3g0fgaTyP9s05rFo7jVnFdS442wzg3dBpF3zJGUDf8Le4+y/PMPw+Z65xhh+LS/EPf4tIWzjic/8zdHlmEoNjwdn7cSICwH5x3IOBjsSEhZHvUM6PFd7MeGs4zsBZLNKx48LAEF+enn4fOx6M5r1wL7h5OFPefwrTfbuJ/FsUd4RE8+BfnsE0Grq2r+rTupTHLSSaR1+YRMhoZ7riyPWTTTz4ZcO9NVQ7AL0HMeL4B2wLfoZz2erGZfAk7t/8DMPvBZeOw4kYUhWgOHN9fzeWTb+Ptc7OcLwcjwffItIV66SohtobMg7viEmEJTpjqHBkQH/vhg9ucpHDmfF8Ni88dA+LXVwwHC/mf86DealqHFwHE2HazFP33QMdXbg/fDAD7d/vZcJjz1NMWFgOpaVcZYrmjXMJ1aq4DmLUzU8xafROHo1ewYNedb8PtzPl9RxmPXofy5yda/rsDnA7U/6cxdNTrONqcR5MxFgvMhvts+FzdO7bH7f3n8K0xpmulGPxCOHVOdYA1Wf4OJY9FYop7RniX7yTJu8x3TOElybO5cWwUHB25MdSZx56/a2apY29BzHC+QPyf2/7fHv7MrA8nl9uO48ATkREREREpAVcUVlZWdn4YWcnJy+fvn0u3EbOTWX5di3PbnjHLiyyaeOCqddzPHXHADzP5k50P+eTvGMJf/luN7mntTmAN+59hTHXXeR7YJeXUnocHJ2dz7DfTQPqLGG6KCrKKS13xLm+ZW7HSyl3bOJ5NFj7QdaMfwvH6HcJ7t4M9dajvKQUznK8ay0VbOZayh2d6x/PxpzNeDeTBsehKd/F47tY8FAWweseb9osneOllDo406xf74pySkvBuaPudS8iIiIiIufnUslO6mrVQREAhz7j1cRXWHq8vo1PHPHsNIJHew3A/3of3Ls6YzTYvWwpxfzjIfZ+k07qdytZ+3MxRfW04tp+BG+aZhGkWxu1qNLsXaRsj2NxSRiJL95+/pswy6Xj+EF2LIxmXkUYic/rsxURERERkcvfJZWd2Gn9QRHAr4dISpjNiz/k1Bv0nDtHAq5+jjdC7sL9ymZtWM5Bae7nfPOLG9ff3BPnizhLRi6C44fJzCjmmv5euOizFRERERGRVuCSy05sfhtBkY059xMWfrqEpcfOfjPbutydRvHsnRMxeTZ5lxMREREREREREeDSzU5+U0FRFfN3u1m3K5alRVm1bmvfOGeCXMN59PYRBPVSQCQiIiIiIiIi5+ZSzU5a513PGmHsNYDwXgMI/7Wcou8zSc1OJ7NoN5nH4Yglx7pRdRsXAgxdoL0Pd7r2xdtrIP7XumDQEjMRERERERERaaV+kzOKRERERERERERa0qWanbRp6QJEREREREREROTSoKBIREREREREREQABUUiIiIiIiIiImKjoEhERERERERERAAFRSIiIiIiIiIiYqOgSEREREREREREAAVFIiIiIiIiIiJio6BIREREREREREQABUUiIiIiIiIiImKjoEhERERERERERAAFRSIiIiIiIiIiYqOgSEREREREREREAAVFIiIiIiIiIiJio6BIREREREREREQABUUiIiIiIiIiImKjoEhERERERERERAAFRSIiIiIiIiIiYqOgSEREREREREREAAVFIiIiIiIiIiJio6BIREREREREREQABUUiIiIiIiIiImKjoEhERERERERERAAFRSIiIiIiIiIiYqOgSEREREREREREAAVFIiIiIiIiIiJio6BIREREREREREQABUUiIiIiIiIiImLTtqULuFjKyo5TfPRnSsylHCsr44SlnIqTJ6ls6cJE5LJwBeDQti3tDI44dehAR6MzLp070aFD+5YuTUREREREpNm0+qDo0A9F/O/wD5w4YaGLS2e6uHTCw/0a2rUz0LZtqz99EWlGJ0+e5MQJC6XHyvjZ/Av5hf+jXTsD13S/GrerXVu6PBERERERkfPWapOSoh+LOZBfgMHREY8e19C1i0tLlyQil7m2bdvi7NwWZ2cnutuCoR9/Kub7//1AQeH/6O3hjmtX/a4REREREZHLV6sMir7O+ZafS8z06d1TAZGIXFBdu7jQtYsLP/5UTN6BgxQf/Zkb+l7X0mWJiIiIiIick1a1mbWlvJy9+zKpBPr3+71CIhG5aLp2caF/v99TCezdl4mlvLylSxIRERERETlrrSYospSXs2//f+jSuRM36n/zRaSF3Nj3Orp07sS+/f9RWCQiIiIiIpedVhMUZf7nG7q5dqFXzx4tXYqI/Mb16tmDbq5dyPzPNy1dioiIiIiIyFlpFUHR1znf4tShg0IiEblk9OrZA6cOHfg659uWLkVERERERKTJLvugqOjHYn4uMWu5mYhccm7sex0/l5gp+rG4pUsRERERERFpkss+KDqQX0Cf3j1bugwRkXr16d2TA/kFLV2GiIiIiIhIk1zWQdGhH4owODrq7mYicsnq2sUFg6Mjh34oaulSREREREREGnVZB0X/O/wD115zdUuXISJyRtdeczX/O/xDS5chIiIiIiLSqMs2KCorO86JE5YLN5vocALjr+vL2JWFzd+2xYz5qBlzWfM3LTZltjG2XMA+KizWPs63k+Zqp7lcAvUUbJrL2CeWk3WGnxHL/uU8ET6XpAvwI9rcunZx4cQJC2Vlx1u6FBERERERkTO6bIOi4qM/08Wlc0uXcU6KNszkFn9/bnk3o6VLqcX8TRqpu9JIL7xEAovzkP6uP7f4+zNtwwVc7pOxwPo5TlvPefXSXO00lxavp4j0tXGkbllA0tfWZ+r7bmb9cwFJu+JY/cUlMWqN6uLSmeKjP7d0GSIiIiIiImd02QZFJeZSOhmvaukyWpXcNeGMDQ9n0W5zS5civ2mumOYn8XFiMjN9rc/U9930eyqZjauTeGekawvVeXY6Ga+ixFza0mWIiIiIiIicUduWLuBcHSsrw8P9mpYuo1kUbFnA6mxwHTSB8H5GwEx6/HKSj4D3/dMZ5lH12JWgPw3lxKYlfLilBM+QKUy9tw/mtDgWxSZwpGsIjzwVRoArYM4gdnkKRXhjmtCDrCWLicnvRujYyYTffvqFdcGWBcR+Yf137pYlzLeMJGKsL0YAcx7JGxJI3JKGxWMkgaaRjOl3ehuWbzYQ80kuFpdAIh61vbcij6R315OFkYEPTSCgO3A4g1Xr15PySR6GW4dgemgkQX2M1kbytzJ/TRZ0C6zu3/xlHDE7i8ArhJlDe1SPl/fIMIzbX2PxLj/mxobh2dhA1+k3dGKYtZ4qFUWkroxjy64UcjsMJDBwJGPu8cLoYP/6Ej5cfYRuj0xk5rX19FFhJjdlPas3bSW1rA+me0Jq2qj6TLoFEnprIR8uWIblgQ+Z27Hm7ScOpxG7OI4tZZ6MDAtjjK9rw20PHsl9o3xxdag5xJy9gVVrkkn5woJn4ECGPlR1jnbfoZCBFMS9TYwlhL/PCcTYlPOqYvf5hA8+wbo33yMFX0InTMDkZTyrWot2xRG7JY3kf4P/iEBMI0Pwq6p1w3qSj8AR43S8/1v/d7Pkn3EkZYMrVT83p3/G9X63vEKYemMuq5YlsKXMnaEPTCB84IUPm5ydOnCw8H8XvB8REREREZHzUnkBfJP73YVotpYd/9pT+euvv164Dg6trhzX27Py4X8UNHvTRxImVvbq7VnZ67X0ysrKysq9r3lW9urtWTku4UjVEZUrx1mfe/UL+8ePVD799FDre3t7Vvbq7V05btb0ykHVjz0re921uDLzZE39vXqHVN5/v2et9zwcl3taTVU1VP8Zt7rySGVlZWVxcuVLd1mfu2Hg0Mpbfaz/vuu19MoTdRs5kVr5ko9nZa/e0ys3l9iey1psrW/kh5U5lZWVlbkfVt5va+Pmu4ZW3lx1Hgm2cf7itdr9n2G8Jk2ZbKv3tcq9DZxP9ZgWrK4c52Pt69YRVf0OrXz1C9tZnEivfOuuqrpGVo4Y6G0956e3VJZUVlZWniyoXDnRu9YY3XXX0Dq1nqjc+5rtOb+hNW1MXF150P4zeXhy5aRBdvVVnfPIiZXjBtX+rJ7+pGogSyo/fdHWts/Ayrtsbfe667XKvcds47RmYuUNttdHjKw5x4VZtb9Dk6YE1tTdpPOyU11rSPXnaP3jV/nSzhNNrjVzid04VdXqM7Fy3aHTv/8NfTdP+4yb+t2aMr3y6Vq1D618K/20b3Oz+/XXXyt3/GvPBe9HREREREQuDxcjOzkXl+3Ss4qTJ2nbtpkmRJVlMD/Yn/EfNbIrbuFWIu8OInpXSy3NSqOgz2vsy8nhs1cCAQvJad2Y91UOB76KIdwJyEsg+Wv79xThP30vB77NYV/sBNyxkDo7juQ62xD5Rebw8WPWfwe9kcqBFSG4AunvTyU2DzwfW82+1CT2/CuWiB6Qu/R5Yr+pU57Bj6EPGIANpKVbO8j9YgsFgF9oIJ4UserVuaQfMxD0RjL7tiWxL3kOQVhInruEpLMc1qSv3YlPzeS/mdPxO+ORZhLnRZF8rA8Rq/ey5/+3d//RVdf3HcefkphvuMEbo4QfmlRrsB4TnScwt4W50zB6gNlJuvYEtSJrZdQj0E1lcynnbJSuh9L1ILVF3UqhLT/cEY5z0Ryb6ukB1BF/kdhJUh1ErEZbSULIhcR8MYH9cW9sZAm5BMIl+Hz8dU9y8/28P9/74Y/74v35fKuq+dWzK5iW1cja72zlHSBsbIIbF7L4Xyt56ZlKqp5bywIgfGIr25ohrFnHsl+GkDefza/uYd+eXSz/7HE38bV13P2jRoLpK3j2pWqqdu7isa8VEP5yOT+u6fPemh2EC6v51e7dPHxTn06W195h0v272ffmHp5dUUpASOXKTdQDvPIQd21uhIL5PPbiTp55bieb78iDxnX846ZGoJk9bYUs+HoF63++k6rKaqpWlACNrPlF37OwatgWg38M8AAADx9JREFULuSZV3fz+prZRJOZV39ey6asak+81n8pBWJs+N5W9iZVax3VDzUCxSz/z2qqKqt5bssyFt9RSHvT/18EA63NjzuJtbXjECX/tZt9e3ax/o68xD1qGHzOpyg9PZ2e7u5hH0eSJEmSTsWIDYpOq+4uwjDGtvtmDRwWNT1NxVcX8WhjAGn9v+VMmFJSTDQN8q8qiP9gUgFXRoFoEYXXAzQd9zS1mcy6Ib71JnrDbObkAWynvjGZ0RrY+VQIFHJ7eTEBQLSEWTcCNFLzP8cfIhww9fNzCYBHX2gAmtm1owEopuyGPIjVUbMdoJzbb8qL/8llM5lZCnRspfb44GkQ0+6cz9QJAUEkOPEbYy+w7SmAKPufe4hVq1ezamsD5AB1O6hthqDwJpYsnEvZ5e1Ub17Nqm8+zKOJP+/qgXcaawmB/FtnMzUKpEWZ+vnZ9B25/vl46DSOBrb8YDWrVj/EtuYoEPLoy32DiLksvqWAaCQg6HuBkvnc/ofxH+T/1W3cDNBUx95mqH/5F4RA0dxyJkcS48+aCcDeF2ppJpepd9zD4i+WELyxlbWrV7JqUy0A4XG5z7w7y5kUjd+3ZObVr9IZ/MVl8Zf5N5VTBvBaLXtjydRaQNGNAVDHilvncPe31/Hz9snMWXjP77eQnayTWVvXz+DPCwJIizJt+syhjSdJkiRJ56gRe0ZRWno63aerqyhawtLKLVA2h7X3zeIOqll/Q5/fNz3JXWX3Ut1WwILNW1haMsQvsykXJTcfaDo+TBpISKwJYBxBVpJDFM9mcd46Vj1VQ/3CSfEv78WzmXYZ8Lt24o0dAdmDJhGnUWfvuI3Ubw/jXS8AOYUU5eTGQ5HGTXzpC/GOlNzCAqYUFzAJqE28NXYgnqxF+4ZS0VwmQbzjBwjb4yHj/vpatu3//duKri2EZO5fAJkfvc7+KKzp6vn9tcdlDbT2Ymz7pzncsbkRcgooKirkmom50HDiLrlk5jWorAsYB8Ah2jth3KC1Rpm1rJLvT3iQLU/toHL9SirXA1mFLNmwhcXFyQ7cR+cQ19b5QxhriLq7u0k7XV2QkiRJkjRMRmxHUWaQQVfXaXyMe6SYpZVbWFAQxjuLftpICHS9senUQqLmp6koK+Mvy1azM1Hu/t/Go4qpl398A83+WGJ/TE+MWOspzucjtez+TeJlWx3bawBKKbosmb8tYPJsgO3U1ffe6yZ2Jw4Wzh/bz71IK2RaeR40/YJtP6uhEph260zyASYUMTkPoIZdvR0e4V4aXgYoJC+nz3VaY7T3xF+2x/qkLkMxYRJFWQAzWfoflVRVVlJVuYVH/n0jj/y0gmkXQe0TK6ntgJvX7uKlqkoe/mY5U/pcIv/yEgDqX26gdxdTc92Oj4Up+VeVAjDpb+5PjFFJ1cYf85OfbuSRuUmkHy83sLs3wPvfWnYCUEB+Lkz6g5sA2PZyPb2fxDtvJGKsS3OJNu9g4+ZGuLaCZ16qpmrD/Syanj/okMnMa6Ba9/QW0lBHNUBWIVdOSKLWnpBYmMvkzy3i4Wd2sW/PLjZ/vRA6Glg11C1gJ7O2UqSrKyQzyEh1GZIkSZJ0QiP2v7ezIhEOd3QyZkyyrS5JSIRFlM1h7Y/WxX/2s3XAKXQS5RZSmN7Ao3UN3LUgxrzP7Kd6fRNQzMyS+BaZ+BfrJ6n/9hy+VP9Zsmsa2HvaPpkGVixaRNP0fPb/chPVQDC7nGn9POQpyM4Dmti5ZTWrwnIW3FbMtJvnk//EOh69+3aiy8rJfX4Tq+qAgnuY82f9t24UlZaTv3o1q1Y3ADMom947WCFlC0tZtXQ7KxYtgjtLaX5iJRs6IJi9iC8UALHJlLGOytdW8qU59Uy76AXq3zjV9qNiyv6+mDXLt3LXV6Isv/NPoOYhlq2vIyxexrOPzSXIzAWa2PX8E+wM8nhn+0o29LlCbmk5ZVnbqXzqG3z57nqmjd3PtsdrCOCjMCR35m3M++ftbPj2Iu7qXsTtE/ez8QcrqW4MmLdhF8sHfSzbVu6e087N07Op3xQ/myj/azOZkgZBaTkL8p5k7dbFfPmCZcwZV8PGH9YBBSy5tZQgeDr+lLk3X6B6eyGTaWDj92oGvTPJzGugWu+6PWReSUDtpk28A0xaODN+VtRgtdLA2jllrGoMmPz177DkeqipiXc2Tc7v/+lj/a3Nj0tibb0y6O0YVoc7OsmKRFJbhCRJkiQNYsR2FGVHx3Awduj0X/ijzqLeH5zqdrM85j3wIPOKo8Se38Sa9U+zN6eEJZsfZF6iqyc6814evqWQgBi1j+8g/Mr9LL/xNMwFgLksX5jNzvXrqGwIid5QwWMrZtDfbIpuWcGCwoDwla2sWbmdvUBQUsHmNXOZnFHH2vuWsuKJBoIb7mHzIwspGuispmtLE2chATfOYlqfbo78W9ZQVVFK7u+eZsV9S1n7fEjRLSuo6q0pOoMla8opyoJY3ZNsC+fz/WWnfo7MpLk/5rGKUqK/XkfFggVUrG8gOr2Cx34yl3yg6K/vZ0lxlL0/W8pt8xawMWsuS0r7XCBnBt99pIKpOSH1T6xjzeMhC9YsY2rfQSKlLK+6n3nFzVSvvJfb/m4l1QeKmfdANctvSCLsur6C736xhUd/uImdbQG5s5ex/m8TZ0NFSlj6yIPMKw6oXb+UipVPUp9RwpLNG1lcmLhvq24il+2sWjCP2+6rZdY/zB98zGTm1W+t9/CtzzWxIVFr0S0Psv5rhYn7MEitaYUs/kni9z+8l9vm3cuaX0eZds8G/u2WvH6H629tHm/QtZViB2OHyI6OSXUZkiRJknRC5x07duzY6b7onsbfcGVBUnubhqyz8wPqXmvgT/94yuBvHtIAdawo+wZ86zSeSRTGiIUB0egZOKDnd1u5Y+pStjGfx96sYHJPGB87mYaGzhD6OSA6bIsRRqIMXn6MysVTuPspmPXALh6+qZ/71xMSi4UE0SjBmTwcPDEuA82jM0YYnLimsDMc/ADtzsRnnTOEz7onJNYdDHyfO2Px3/f3hp6QWAdDWmNJzeuVlXx6zjooXcFL68vJPZVaIf5vopPk18EAa/NjUrW2BvHfL+6i+NpCIpHRqS5FkiRJ0lngTGQnQzFit55FIqPJzAxoaT3A2IsvGoYBiln6TPXpvWaQTMgyTNKSDIlgwC/iQU500KdhNT9+L199aEf8qWoFC1kwY4CQLW2IIcqpGmzcyOBzHDRMSVwn6ft9vLSA6IkCjkh04A6ZtIDoEHPNpObV33hDrRVO/t9EMjWmam2dQEvrATIzA0MiSZIkSWe9ERsUAVwyYTzvvvf+8ARFI11aNuOuLaSI/s98GR4he95qhEgeRTNnsvRbC5l8dn1f16lKy40/xW1CdqorGVHefe99LpkwPtVlSJIkSdKgRuzWs14v7foVV1yeb1gk6azU0nqAN996hz+acl2qS5EkSZJ0Fjlbt56N2MOse336snwa972d6jIkqV+N+97m05flp7oMSZIkSUrKiA+KcsdexIXZUV7f82aqS5Gkj3l9z5tcmB0ld6wdj5IkSZJGhhEfFAFcdeUVdHR28tbbTakuRZIAeOvtJjo6O7nqyitSXYokSZIkJe2cCIoArrn6M+xvbjUskpRyb73dxP7mVq65+jOpLkWSJEmSTso5ExQFGRlcd+3VtLYddBuapJR5fc+btLYd5LprrybIyEh1OZIkSZJ0Us6ZoAjiYdGU667hPODFV16lpfVAqkuS9AnR0nqAF195lfOAKdddY0gkSZIkaURKT3UBw+GqK6+guSX+SOp333ufSy8Zz9iLPUxW0unX0nqAd997n/DIEa64/FMeXC1JkiRpRDsngyKIPw0td+xF/Pb9Zn7T9B5v7N3HxRflcGH0AsZkRcjMDEhPP2enL2kYdHd309UVcrijk4OxQ7QeaCMzM+CSCeOZOD431eVJkiRJ0ik755OSieNzmTg+l87ODzjQdpDWAwd5u+k9usIj9HR3cyzVBUoaEc4D0tLTyQwyyIpEyI6O4VOXTiQSGZ3q0iRJkiTptDnng6JekchoIpHR5F2a6kokSZIkSZLOTufUYdaSJEmSJEkaOoMiSZIkSZIkAQZFkiRJkiRJSjAokiRJkiRJEmBQJEmSJEmSpASDIkmSJEmSJAEGRZIkSZIkSUowKJIkSZIkSRJgUCRJkiRJkqSEYQmK0tJG0dNzdDguLUmSJEmSNKL19BwlLe3s7N0ZlqrS09P5sPvD4bi0JEmSJEnSiPZh94ekp6enuox+DUtQFAQZhF1HhuPSkiRJkiRJI1rYdYQgyEh1Gf0alqAokplJ5wddw3FpSZIkSZKkEa3zgy4imZmpLqNfwxIUZWWN5nBHp+cUSZIkSZIk9dHTc5TDHZ1kZY1OdSn9GpagaNSoUUSjY2iPHRqOy0uSJEmSJI1I7bFDRKNjGDXqE3SYNcCF2VHaDsbsKpIkSZIkSSLeTdR2MMaF2dFUlzKgYQuKgozzyY6OoaW1bbiGkCRJkiRJGjFaWtvIjo4hyDg/1aUMaFj7nMZenEN45Ajt7W5BkyRJkiRJn1zt7YcIjxxh7MU5qS7lhIZ9Q9z4cWNpbTvIocMdwz2UJEmSJEnSWefQ4Q5a2w4yftzYVJcyqGEPioKM85k4YRzNLQfsLJIkSZIkSZ8o7e2HaG45wMQJ487qLWe9zjt27NixMzFQeORD3t/fQpCRwdiLc0hLOztP95YkSZIkSTpVPT1HaWltIzxyhPHjxo6IkAjOYFDUq6W1jfbYYXIujJIdvcDASJIkSZIknTN6eo7SHjtE28EY2dExZ/2ZRMc740ERxLuLDrbHiMUOMyYrQmR0JkFmBuenn29wJEmSJEmSRoyenqN82P0hYdcROj/o4nBHJ9HoGC7Mjo6YLqK+UhIU9Tp69CgdHR/Q2dVFGB6hu7ubnp6jqSpHkiRJkiTppKSljSI9PZ0gyCCSmUlW1mhGjRq5TTApDYokSZIkSZJ09hi5EZckSZIkSZJOK4MiSZIkSZIkAQZFkiRJkiRJSjAokiRJkiRJEmBQJEmSJEmSpASDIkmSJEmSJAEGRZIkSZIkSUowKJIkSZIkSRJgUCRJkiRJkqQEgyJJkiRJkiQBBkWSJEmSJElKMCiSJEmSJEkSYFAkSZIkSZKkBIMiSZIkSZIkAQZFkiRJkiRJSvg/U4amrpDH2sEAAAAASUVORK5CYII="
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![image.png](attachment:image.png)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Submitting on Kaggle gives us a accuracy of 0.98585"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "6fcpbYGFi356"
   },
   "source": [
    "## Conclusion\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### On SVM"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "-\tThere was no need of any outlier treatment(we have to consider all the values to predict a label). There was no need for dummy variables or any conversion to categorical variables. Scaling or Normalization may be required before PCA. \n",
    "-\tSince we have 784 features explaining a label, it was necessary that we bring in PCA to summarize the content and explain the variance using Principal Components\n",
    "-\tThe model was trained on two different kernel types – RBF and Linear. RBF was giving the highest accuracy explaining 86% variance in the data set. \n",
    "-\tWith n_components= 0.86, C = 10 and gamma = 0.01, Training samples have significant impact on the model – giving an accuracy of ~99%\n",
    "-\tAccuracy when predicted labels (predicted on test data) are uploaded to Kaggle gives us around 98%\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### On CNN"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- Again, there no need of any outlier treatment or pre-processing of the dataset other than scaling or normalization.\n",
    "- We have choosen Keras Sequential Model\n",
    "- This model gives us an accuracy of 98.5% on the test data when uploaded to kaggle"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Thus, using CNN it gives us the highest accuracy on test data set = 98%"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "name": "MNSIT(Vector Victors) (2).ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}