Loading…
Glossary term
A model's ability to perform well on new data it has not seen before.
Generalization is performing well on data the model has never seen. It is the real goal of machine learning: not to recall the training set, but to handle new cases.
It is measured by holding out data during training and scoring the model on it afterwards.
A model's value comes entirely from how well it generalizes. Benchmarks and test sets exist to measure exactly that.