Suppose the goal is to build a model that estimates response variable, Y, as a function of p predictors, X1, ..., Xp. Assume k is a positive integer less than p. Best subsets regression finds the k-variable linear regression model with the highest R2 by:

a) using the correlation between predictors and response.

b) starting from the best model with k-1 predictors and finding the next best variable to add.

c) fitting all k models that use the subset of k predictors.

d) fitting the model with the entire set of p predictors, and the removing the p - k predictors with the lowest correlation with the response.

e) None of the answers are correct.