machine learning - Combining training, validation and test datasets -
is possible train model based on training , validation data sets.basically end combining both of them create new model. , combined model use classify of data in test dataset.
this done. assuming know how transfer hyperparameters, fit model on train data, select hyperparameters based on score on valid one. when combine train + valid bigger dataset, "optimal hyperparameters" might different ones selected before. in general - yes, done, might more tricky expect (especially if method highly stochastic, non deterministic, etc.).
Comments
Post a Comment