Open zar1 opened 9 years ago
Currently, we hardcode model.cross_validation.CrossValidationScore to use sklearn.cross_validation.KFold. We should be able to wrap around arbitrary sklearn cross-validators
In particular, cross-validation by time. Relatedly, we should be able to train/test split by time.
Currently, we hardcode model.cross_validation.CrossValidationScore to use sklearn.cross_validation.KFold. We should be able to wrap around arbitrary sklearn cross-validators