STAT 425 Week 10 - Shrinkage Methods for Regularized Regression - Part 2

Introduces penalized least squares methodology for stabilizing linear regression with correlated or high dimensional predictors. Two key methods are Ridge Regression, using an L2 penalty, and Lasso regression, using an L1 penalty. Ridge regression is an effective way to stabilize models with correlated predictors. Lasso regression has a useful variable selection property for situations where the the model is expected to be sparse, i.e., to depend on only a few of the many potential predictor variables.