BostonDataSet_LinearRegressionModel

Analysis of Boston data set – sklearn.linear_model.LinearRegression – fit, predict, score – StatsModels • STEPS- (1) Find the best model with <= 3 predictors in terms of RMSE to predict medv in the whole Boston data set •

1. Loading of boston dataset
2. Checking boston housing stats
3. Spliting the data in training and testing data
4. Model prediction and RMSE verification
5. Finding correlation among the features and the price
6. Eliminating the features with higher p values
7. Finallizing 3 most enfluencing predictors- RM, PTRATIO, LSTAT
8. Creating new model with these predictors
9. Spliting the data in trained and test sets
10. Normalizing and scaling the data
11. Fitting the model
12. Predicting the test set results
13. Verifying the R2 and RMSE

(2) Find the best mode with <= 3 predictors including log, square, cubic transformation in the whole data set . STEPS-

1. Loading the boston dataset
2. Spliting the data
3. Model prediction and RMSE, Rsq verification
4. RMSE , Rsq verification with 3 most correlated predictors
5. Building interaction model and verifying RMSE
6. Building a model with square polynomial transformation
7. Building a modelwith log beta
8. Building a model with square root transformations
9. Comparing the models in terms of Rsq and RMSE.