overfitting.

The adaptation of a model too closely to the particular training data, including noise or incidental features, so that generalization suffers.

Use it for

Words that belong together

• risk of overfitting
• overfitting to training data

On the page

The risk of overfitting increased when the model had many adjustable parameters and little data.

Overfitting to training data produced excellent internal scores but weak performance on the reserved cases.

Usage notes

Primary category: science / research / methodology. Forms: Usually uncountable. Legacy lexical fields mirror sense 1 only. Use numbered sense fields for all meanings. Established modern English lexical item. No separate regional spelling or grammatical inflection entry. Writing functions: method specification, measurement critique, model interpretation, qualification. Disciplines: research methods, statistics, experimental design. Commonness: moderately common.