Building Regression Model

[aditbala99]

#Building Regression model
library(caTools)
set.seed(123)
split = sample.split(datan$median_house_value, SplitRatio = 0.9)
training_set = subset(datan, split == TRUE)
test_set = subset(datan, split == FALSE)
m <- lm(formula = median_house_value ~ ., data = training_set)
summary(m)

Call:
lm(formula = median_house_value ~ ., data = training_set)

Residuals:
Min 1Q Median 3Q Max
-365334 -44799 -8876 33157 516625

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -17713.419 3007.209 -5.890 3.92e-09 ***
housing_median_age 1403.913 46.780 30.011 < 2e-16 ***
total_rooms -16.570 1.225 -13.529 < 2e-16 ***
total_bedrooms 125.034 8.350 14.973 < 2e-16 ***
population -64.708 1.588 -40.739 < 2e-16 ***
households 157.760 9.630 16.382 < 2e-16 ***
median_income 51145.454 470.785 108.639 < 2e-16 ***
ocean_proximity.INLAND -63853.921 1348.467 -47.353 < 2e-16 ***
ocean_proximity.ISLAND 171468.785 31044.200 5.523 3.37e-08 ***
ocean_proximity.NEAR BAY -1717.301 1767.588 -0.972 0.331
ocean_proximity.NEAR OCEAN 11860.808 1620.804 7.318 2.62e-13 ***

Signif. codes: 0 ‘’ 0.001 ‘’ 0.01 ‘’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 69340 on 18673 degrees of freedom
Multiple R-squared: 0.6453, Adjusted R-squared: 0.6451
F-statistic: 3397 on 10 and 18673 DF, p-value: < 2.2e-16

Predicting the Test set results

y_pred = predict(m, newdata = test_set)
MSE <- mean((y_pred - test_set$median_house_value)^2)
MSE
[1] 4247739708
totalss <- sum((test_set$median_house_value - mean(test_set$median_house_value))^2)
totalss
[1] 2.08637e+13

Regression and Residual Sum of the Squered.

regss <- sum((y_pred - mean(test_set$median_house_value))^2)
regss
[1] 1.476199e+13
resiss <- sum((test_set$median_house_value - y_pred)^2)
resiss
[1] 8.308579e+12

Calulate R squared.

R2 <- regss/totalss
R2
[1] 0.7075441