Binder Link :
Description: This project analyzes the Broward County dataset to identify the key factors that contribute to the likelihood of future criminal activity. Machine learning techniques are used to process various features and predict which features increase the chances of committing crimes. This project aims to predict future crimes using machine learning techniques based on the Broward County dataset. The scientific motivation behind this analysis is to understand the factors that contribute to criminal behavior and use this knowledge to develop effective crime prevention strategies.
Analysis Steps
-
Data Preparation: Perform necessary data cleaning, preprocessing, and feature engineering techniques to prepare the dataset for analysis.
-
Exploratory Data Analysis: Conduct exploratory analysis to gain insights into the dataset, visualize relationships between variables, and identify potential patterns.
-
Feature Selection: Use feature importance techniques, such as Random Forest, to identify the most influential features for predicting future crimes.
-
Model Training: Train a classification model, such as Random Forest Classifier, using the selected features and the target variable (crime prediction).
-
Model Evaluation: Evaluate the performance of the trained model using appropriate evaluation metrics and cross-validation techniques.
-
Interpretation: Interpret the model results and analyze the findings to understand the significant factors influencing crime prediction.
Results
The analysis revealed that age, prior misdemeanor charges, and the total count of prior charges were the most important features for predicting future crimes. Younger individuals and those with a higher number of prior offenses were found to be at a higher risk of reoffending.
Please refer to the complete analysis report in Main.ipynb for detailed findings and insights.