With ridge regression, we will have all the eight features in the model, so this will be an intriguing comparison with the best subsets model. The package that we will use and is in fact already loaded, is `glmnet`. The package requires that the input features are in a matrix instead of a data frame and for ridge regression, we can follow the command sequence of `glmnet(x = our input matrix, y = our response, family = the distribution, alpha=0)`. The syntax for alpha relates to `0` for ridge regression and `1` for doing LASSO.

To get the `train` set ready for use in `glmnet` is actually quite easy by using `as.matrix()` for the inputs and creating a vector for the response, as follows:

> x <- as.matrix(train[, 1:8])> y <- train[, 9] ...