- (statistics) To use a statistical model that has too many parameters relative to the size of the sample leading to a good fit with the sample data but a poor fit with new data.
- Antonym: underfit
- 2004, Mark T.D. Cronin, Predicting Chemical Toxicity and Fate, CRC Press (→ISBN), page 169:
- In all modeling techniques, and neural networks in particular, care must be taken not to overtrain or overfit the model. If possible, models should be interpreted in terms of their mechanistic meaning.
- 2007, Kenneth P. Burnham, David R. Anderson, Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach, Springer Science & Business Media (→ISBN), page 250:
- If your strategy is to always fit and use the global model, you will probably overfit the model (i.e., include unnecessary variables).
statistics: to use a statistical model that has too many parameters