13.6 Augmenting features. You are fitting a regression model y = $ 8 + v to data, using least squares without regularization, computing the model coefficients / and v via the QR factorization. A friend suggests adding a new feature, which is the average of the original features. (That is, he suggests using the new feature vector a = (@, avg(@)).) He explains that by adding this new feature, you might end up with a better model. (Of course, you would test the new model using validation.) Choose one of the following, and then explain why below. . This is a good idea, and it's worth a try. 42