Skip to main content
All terms
Evaluation

Cross-Validation

Testing a model by training and checking it on different slices of the data in turn.

Definition

Cross-validation estimates how well a model will perform on new data by splitting the data into several parts, then training on some and testing on the held-out part, rotating through all of them. Averaging the results gives a more reliable score than a single train/test split, especially with limited data.