-
Notifications
You must be signed in to change notification settings - Fork 0
/
Predicting Housing Price.Rmd
242 lines (223 loc) · 10.1 KB
/
Predicting Housing Price.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
---
title: "Predicting Ames, Iowa Housing Price"
author: "Alex Navarro"
date: "10/27/2021"
output:
html_document:
df_print: paged
pdf_document: default
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
# Load Packages:
library(MASS)
library(ggplot2)
library(markdown)
library(knitr)
library(RColorBrewer)
library(Sleuth2)
# Read data into system:
data <- read.csv("AmesHousing.csv")
```
# Exploratory Data Analysis
```{r, echo=TRUE}
# Set Seed to Save Models:
set.seed(1)
# check head of dataframe:
head(data, n = 10)
# Look at structure of dataframe:
str(data)
# Look at missing values in dataframe:
summary(data)
```
```{r, echo=TRUE}
# Remove Missing Values from Data:
data2= na.omit(data)
# check structure of dataframe:
str(data2)
# Split Training Set 70/30
train <- sample(2258,1800)
test <- (c(1:2258)[-train])
# Create a data frame with continuous variables only:
num.ames =data.frame(data2[,c(2,3,14:17,23,31,33:35,40:49,51,53,56,58,59,63:68,70,71,74)])
# Checking Data Correlation and Distribution:
plot(SalePrice ~., data = num.ames, subset = train)
```
# Modeling:
```{r, echo=TRUE}
# Create First Model:
fit <- lm(SalePrice ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + TotRms.AbvGrd +
Garage.Yr.Blt + Wood.Deck.SF + Open.Porch.SF, data = num.ames, subset = train)
# Return model summary of first model:
summary(fit)
# Create Second Model:
fit2 <- lm(SalePrice ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + TotRms.AbvGrd +
Garage.Yr.Blt + Wood.Deck.SF, data = num.ames, subset = train)
# Return model summary of second model:
summary(fit2)
# Create Third Model:
fit3 <- lm(SalePrice ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Garage.Yr.Blt + Wood.Deck.SF, data = num.ames, subset = train)
# Return model summary of third model:
summary(fit3)
```
# Check Model Diagnostics:
```{r, cache=TRUE}
# Plot Fitted vs Residuals:
plot(fit3$res~fit3$fitted, main = "Fitted vs Residuals")
# Check normality of model:
hist(fit3$res, main = "Normality Test",
col = c("blue", "red", "green"))
# Plot qq-plot:
qqnorm((fit3$res))
# add reference line:
qqline(fit3$res)
# Compute Shapiro-Wilk Test for Normality Check:
shapiro.test(fit3$res)
```
# Boxcox Transformation:
```{r, cache=TRUE}
# Run boxcox transformation to help normalize data:
boxcox(SalePrice~Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + TotRms.AbvGrd +
Garage.Yr.Blt + Wood.Deck.SF, data = num.ames)
# Create new variable that is the log of SalesPrice:
num.ames$SalePriceLog <- log(num.ames$SalePrice)
# Create new model using SalesPriceLog for the dependent variable:
fit4 <- lm(SalePriceLog~Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + TotRms.AbvGrd +
Garage.Yr.Blt + Wood.Deck.SF, data = num.ames)
# Plot Fitted vs Residual Values:
plot(fit4$res~fit4$fitted, main = "Diagnostic Check Model 4")
```
# Determine Categorical Variables For Model:
```{r,cache=TRUE}
# Which Categorical Variables Should be Included:
# Create Model One:
model1 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF, data = data2, subset = train)
# Create Model Two:
model2 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF + Street, data = data2, subset = train)
# Run Anova:
anova(model1,model2)
# Create Model One:
model1 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF, data = data2, subset = train)
# Create Model Two:
model2 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF + Street + Lot.Config, data = data2, subset = train)
# Execute Anova:
anova(model1,model2)
# Create Model One:
model1 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF, data = data2, subset = train)
# Create Model Two:
model2 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF + Street + Lot.Config + Foundation, data = data2, subset = train)
# Execute Anova:
anova(model1,model2)
# Create Model 1:
model1 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF, data = data2, subset = train)
model2 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF + Street + Lot.Config + Foundation + Bsmt.Exposure, data = data2, subset = train)
anova(model1,model2)
# Create Model 1:
model1 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF, data = data2, subset = train)
# Create Model 2:
model2 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF + Street + Lot.Config + Foundation + Bsmt.Exposure + Heating.QC, data = data2, subset = train)
# Execute Anova:
anova(model1,model2)
# Create Model 1:
model1 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF, data = data2, subset = train)
# Create Model 2:
model2 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF + Street + Lot.Config + Foundation + Bsmt.Exposure + Heating.QC + Central.Air, data = data2, subset = train)
# Execute Anova:
anova(model1,model2)
# Create Final Model:
model1 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF, data = data2, subset = train)
# Create Final Second Model:
model2 = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add + BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area + Wood.Deck.SF + Open.Porch.SF + Street + Lot.Config + Foundation + Bsmt.Exposure + Heating.QC + Central.Air + Functional, data = data2, subset = train)
# Execute Anova:
anova(model1,model2)
# New Model: Both Numeric and Non-Numeric:
fitnew = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add +
BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area +
Wood.Deck.SF + Open.Porch.SF + Street + Lot.Config + Foundation +
Bsmt.Exposure + Heating.QC, data = data2, subset = train)
# Run Summary of Model:
summary(fitnew)
# Remove Final Variables and Rerun Model:
fitnew = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add +
BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area +
Wood.Deck.SF + Open.Porch.SF+ Foundation +
Bsmt.Exposure + Heating.QC, data = data2, subset = train)
# Run Summary of Model:
summary(fitnew)
```
# Check Diagnostics:
```{r, echo = TRUE}
plot(fitnew$res~fitnew$fitted, main = "Residuals vs Fitted")
# Check Skewedness of Model:
hist(fitnew$res, main = "Diagnostic Check Model Histogram",
col = c("blue", "red", "green"))
# Normality Test:
qqnorm((fitnew$res))
# Reference Line:
qqline(fitnew$res)
# Shapiro-Wilkins Test:
shapiro.test(fitnew$res)
```
# Run Test Data:
```{r, echo = TRUE}
#Final Model:
final_final = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add +
BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area +
Wood.Deck.SF + Open.Porch.SF+ Foundation +
Bsmt.Exposure + Heating.QC, data = data2, subset = test)
# Summary Final Model:
summary(final_final)
# Remove Foundation:
#Final Model:
final_final = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add +
BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area +
Wood.Deck.SF + Open.Porch.SF+
Bsmt.Exposure + Heating.QC, data = data2, subset = test)
# Summary Final Model:
summary(final_final)
# Remove Open.Porch:
#Final Model:
final_final = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add +
BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area +
Wood.Deck.SF + Foundation +
Bsmt.Exposure + Heating.QC, data = data2, subset = test)
# Summary Final Model:
summary(final_final)
# Remove Foundation:
#Final Model:
final_final = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add +
BsmtFin.SF.1 + Total.Bsmt.SF + X1st.Flr.SF + Gr.Liv.Area +
Wood.Deck.SF +
Bsmt.Exposure + Heating.QC, data = data2, subset = test)
# Summary Final Model:
summary(final_final)
# Remove Total.Bsmt.SF:
#Final Model:
final_final = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add +
BsmtFin.SF.1 + X1st.Flr.SF + Gr.Liv.Area +
Wood.Deck.SF +
Bsmt.Exposure + Heating.QC, data = data2, subset = test)
# Summary Final Model:
summary(final_final)
```
# Final Check Diagnostics:
```{r, echo = TRUE}
plot(final_final$res~final_final$fitted, main = "Residuals vs Fitted")
# Check Skewedness of Model:
hist(final_final$res, main = "Diagnostic Check Model Histogram",
col = c("blue", "red", "green"))
# Normality Test:
qqnorm((final_final$res))
# Reference Line:
qqline(final_final$res)
# Shapiro-Wilkins Test:
shapiro.test(final_final$res)
ameshousing.predict=read.csv("AmesHousing_predict.csv")
library(Sleuth2)
price = lm(num.ames$SalePriceLog ~ Overall.Qual + Year.Built + Year.Remod.Add +
BsmtFin.SF.1 + X1st.Flr.SF + Gr.Liv.Area +
Wood.Deck.SF +
Bsmt.Exposure + Heating.QC, data = data2, subset = test)
newdata=ameshousing.predict
predict(price, newdata=ameshousing.predict, interval="prediction")git
```