Multiple Linear Regression : Real Estate

REAL ESTATE

Objective : Create a model on the real estate that predicts the price of the sold house per unit area.The data contains following features that may or may not impact the selling of a house.

Dataset: RealEstate.csv

Rcode:RealEstate.R

Lets Begin!!

Step 1: Load required libraries and data to R studio

Step 2: Analyze data

Correlation lies between 1 and -1(positive correlation and negative correlation)

Step 3: Outlier Detection

We need to check for each variable.

Step 4: Data Partition

Splitting the data into train and test data(75:25).

We will train the  train.real data and test the model on test.real.

Step 5: Model

lm() is used for linear regression.

step() is used for getting better model in terms of more correlated variables.

if vif < 5 than no multicollinearity if vif > 5 than multicollinearity is present

Here P value is more than 0.05 hence we fail to reject H0, but we do not have evidence to believe full model is better.

Step 6: Plotting model and residual

Plot 1: Independent variables VS Residuals(error)  

Plot 2: Check heteroscedasticity(Residual vs Price)

Plot 3: Model

For durbinWatson Test, rho !=0 means there is correlation between Residuals and Regression analysis.

Step 7: Prediction time!!

We got the predicted values and saved the Observed and Predicted  values in a file.

Accuracy 

Step 8: Visualization

Happy Learning!!!!