A bookmobile arrival model grows huge coefficients on a couple of rare street names and then swings wildly on the next week's routes. Which regularization technique fits?
Select an answer to reveal the explanation.
Short Explanation
Think of huge coefficients on a couple of rare street names, then wild swings next week. L2 regularization, weight decay. Dropout is a neural-net layer drop, not that weight penalty.
Full Explanation
L2 regularization, also called weight decay, penalizes large weights so rare features cannot dominate. Dropout is a neural-net regularizer, not the weight-penalty for huge coefficients. Fraud Detector and dropping regularization do not fix that swing.