Glossary Entry

Overfitting

A failure mode where a model matches the training data too closely and performs worse on new, unseen examples.

Training Generalization

Seed source: Google ML Glossary

An overfit model learns accidental quirks and noise instead of only the reusable signal. It can look impressive on training data while still failing when the inputs shift even modestly.

Several posts here touch this risk directly, especially around tuning, encoding tricks, and evaluation. It is one of the main reasons validation strategy and regularization matter so much in practice.