overfitting
Say it: oh-vuh-FIT-ing /ˌəʊvəˈfɪtɪŋ/ · noun
What “overfitting” means
Learning old examples so closely that new ones cause mistakes.
More precisely: When a model learns its training data too closely, including the noise, and fails on new cases.
Example: “Copying your one perfect race day exactly, down to the lucky socks, is overfitting.”
When to use it
In machine learning or analytical discussion, when a model matches familiar data brilliantly but performs badly on fresh examples.
Similar words
over-tuning, memorising noise
Quick check
A model gets every practice question right but fails new questions because it memorised accidental details. What has happened?
overfitting or underfitting?
Show the answer
overfitting
Learn 5 new words a day in Kinquire. Each with a picture, how to say it, an example and a quick check, then spaced review so you actually keep them. Free to start.
Learn “overfitting”