Linked from

The 24 pages that link to Overfitting, each with the reason it gives.

• Machine learningRelated: A model can memorize examples rather than learn patterns that generalize.

• Regression analysisRelated: Excessive model complexity can make regression predictions unreliable outside the fitting data.

• Statistical modelRelated: Excessive flexibility can make a fitted model describe the sample rather than the underlying relationship.

• Neural networkRelated: Excessive fitting can make a network unreliable beyond its training examples.

• Mathematical modelingRelated: A model can match its calibration data closely yet make unreliable predictions elsewhere.

• Mathematical modelRelated: A model can match observed data closely while misrepresenting the wider system.

• Model selectionRelated: It explains why the best-fitting candidate need not be the best model.

• Training dataRelated: Memorizing peculiarities of the training examples can undermine performance on unseen data.

• Data miningRelated: Searching many candidate patterns can produce convincing results that fail outside the dataset.

• Economic modelRelated: A model that fits past observations closely can still perform poorly outside its estimation sample.

• No-free-lunch theoremRelated: Without assumptions about data structure, observed optimization success need not transfer.

• ExtrapolationRelated: A model that captures noise rather than a stable pattern extrapolates unreliably.

• Minimum description lengthRelated: The model-description penalty discourages fitting noise with unnecessary complexity.

• Coefficient of determinationRelated: Adding predictors can increase in-sample R² even when predictive performance worsens.

• Nonlinear least squaresRelated: Flexible nonlinear models can achieve low fitting error without generalizing well.

• Parametric statisticsRelated: Adding parameters can improve in-sample fit while weakening generalization.

• Generative AIRelated: Overfitting can make generated outputs repeat training examples rather than generalize.

• Scientific modellingRelated: Overly flexible models can mistake noise for stable structure.