Generalization
Also: generalisation, generalization error, generalization gap
The ability of a model to make accurate predictions on unseen inputs. It is estimated with held-out validation or test data, and the difference between training and test performance is called the generalization gap.
Example
A classifier scoring 98% on its training set and 85% on the untouched test set has a generalization gap of 13 percentage points; the 85% is the honest estimate.