Overfitting

noun

say it: “OH-ver-fit-ing” /ˈoʊvərˌfɪtɪŋ/

When a model fits the examples it learned from so closely, quirks and all, that it guesses worse on new examples.

What is overfitting?

Overfitting is when a model learns its examples' random ups and downs so closely that it scores well on them and worse on new examples, like memorizing last year's practice test instead of learning the subject. A much worse score on test data than on training data is the sign.

In a sentence

  • “Overfitting showed up as a perfect training score and a poor test score.”
  • “Using fewer, simpler features cut down the overfitting.”

Level: Journeyman · More AI Basics terms at this level

Updated