hard20 pts · first answer countsmachine-learningensemblesstacking

Training a stacking meta-model

In stacking, a meta-model learns to combine the predictions of several base models. Why must the meta-model be trained on out-of-fold base predictions?

A.In-sample predictions look too good, so overfit models get trustedB.Out-of-fold predictions are much faster to compute than othersC.The meta-model cannot read probabilities from in-sample rowsD.The base models are retrained by the meta-model at each fold

Discussion 0

Join free to post in the discussion.

Training a stacking meta-model

Discussion 0

Related questions