Intermediate · 35 minYou are hereRead
Bias and variance of a model
Refit one model on many training sets from one curve: a too-simple model misses the same way each time, a too-flexible one changes each time.
Classical machine learning
Generalisation and evaluation
A model is judged by data it has not seen, not by how well it fits its training set. This topic separates bias from variance, holds data out with cross-validation, reads learning curves, and measures a classifier's two kinds of mistake.
Start with Bias and variance of a model0 of 4 read
Lessons on this topic
Related topics