Regression: Choosing and Managing Models
This chapter generalizes our understanding of regression in a number of ways. The previous chapter showed we could at least reduce training error, and quite likely improve predictions, by inserting new independent variables into a regression. The difficulty was knowing when to stop. In Sect. 11.1, I will describe some methods to search a family of models (equivalently, a set of subsets of independent variables) to find a good model. In the previous chapter, we saw how to find outlying points and remove them. In Sect. 11.2, I will describe methods to compute a regression that is largely unaffected by outliers. The resulting methods are powerful, but fairly intricate.