Two data scientists are discussing a strategy to select a subset of predictors for a model with n = 5,000 observations and p = 400 predictors. The rst suggests that they perform a forward stepwise selection procedure starting with a null model. Of these resulting models, they would nally choose the one with the smallest RSS. The second objects, saying that forward stepwise selection is a greedy algorithm and is unlikely to nd the true optimal model. Therefore, they should instead use a best subset algorithm. Do you agree with either of the data scientists? Explain your answer.