Variable selection in regression analysis is considered in Section 1.7. Criteria for selection include the principle of parsimony, the adjusted R2 a, the AIC, BIC, and Cp statistics. Variable...

Variable selection in regression analysis is considered in Section 1.7. Criteria for selection include the principle of parsimony, the adjusted R2

a, the AIC,

BIC, and Cp statistics. Variable selection strategies include backward elimination, forward selection, and stepwise selection, the last one being preferred.

Illustration of multiple regression and variable selection with air pollution

data is given in Section 1.8.