11.7 STEPWISE REGRESSION
In this small example, we had only three predictors. But, most data science projects use dozens if not hundreds of predictors. We therefore need a method to ease the selection of the best regression model. This method is called stepwise regression. In stepwise regression, helpful predictors are entered into the model one at a time, starting with the most helpful predictor. Because of multicollinearity or other effects, when several helpful variables are entered, one of them may no longer be considered helpful any more, and should be dropped. For this reason, stepwise regression adds the most helpful predictors into the model one at a time and then checks to see if they all still belong. Finally, the stepwise algorithm can find no further helpful predictors and converges to a final model.
The application of stepwise regression (not shown) to the clothing_sales_training and clothing_sales_test data sets converged on the final models displayed in Figures 11.3 and...