Suppоse thаt yоu wаnt tо predict the vаlues of a numerical response, Y, using a single numerical predictor, x. You expect that there is very nearly a linear relationship between E(Y|x) and x, and the error term variance is relatively large. If the sample size of the training data is small, and the focus is on obtaining accurate predictions of future values of Y for given new values of x, what should you do?