The difference between using predict() on BasetableVAL and BasetableTEST During the logistic regression video, at the very end, you switch from using BasetableVAL to BasetableTEST. Why did you do that? Is that the
cross-validation of lambda or is the cross-validation when you use the for() loop to run through every value of lambda?
Answers and follow-up questions Answer or follow-up question 1
VAL is for anything related to tuning parameters (lambda in this case). Tuning means predicting multiple times on VAL each time with a
different parameter value and then selecting the best model (and corresponding lambda). At the end we want to know the final performance of
the best model and we need to determine that on an untouched dataset (i.e., test).
Michel BallingsSign in to be able to add an answer or mark this question as resolved.