Linear_regression

A small program to implement multiple linear regression with Sklearn library and on my own.

Input data:

For this example I used a data set on the cost of houses for 1500 rows

Columns:

Data processing:

• One-hot encoding for categorial features
• Divide the sample randomly into train and test data (test - 25% of the data)
• Fill nan - in numerical features from the sample train
• Standard scaling by Train sample

Results:

sklearn:

Train mse: 19306.85068414516
Test mse: 3.3287891454200028e+16

Self-written gradient descent:

Train MSE GD_Regressor: 197151.96183913611
Test MSE GD_Regressor: 198865.79451044145