You are comparing a few supervised learning models and trying to understand why some underfit while others overfit.
What is the bias-variance trade-off?
