Main Conclusions of the Chapters 4–17

This chapter gives a summary of the regularized regressions from the data examples applied in this edition. The regularized regressions included either Ridge, or Lasso or Elastic Net coefficients. Ridge tends to retain all of the x-variables in a data set. Lasso only selects the computations from the largest x-variables and pushes out the rest. Elastic Net selects some variables and retains them. Lasso is best suitable for data with a limited number of strong predictors. Elastic Net is much like Lasso, but will perform better, if many predictors are in the data.