What overfitting is, what regularization does about it, and how to choose without fooling yourself.
8h · 11 lessons · 1 challenge
Before this: Classification & Logistic Regression
Overfitting is the failure every model can have and the one beginners cannot see. This module names it, shows the three ways to address it, and derives regularization as a penalty on capacity.
The second half is model selection: a validation set that stays honest, cross-validation and the ways it leaks, and the search strategies that spend your compute well.
Twelve learning curves. Diagnose each one and prescribe the single most valuable next action.
Worth 400 XP.
