Basics

In this part of the book, we introduce the concept of supervised learning and cover key concepts you need for building a sparse prediction model for real-valued responses.

• In 1  What is supervised learning?, we explain what supervised learning is and describe ordinary least squares (OLS) model as a basic example.

• In 2  Regularized regression, we introduce the lasso and the elastic net as sparse alternatives of OLS, and walk through how to use the glmnet package in R to build such models.

• In 3  Details on the lambda sequence, we discuss the role that the regularization parameter \(\lambda\) plays in tuning the lasso and elastic net models.

• Building on that, 4  Cross-validation introduces cross-validation (CV) as a recommended strategy for deciding on the value of \(\lambda\) for your lasso/elastic net model.

• Finally, in 5  Standardization we discuss the effect that standardizing variables has on the prediction model.