tabensemb.model.AutoGluon._train_data_preprocess#

method

AutoGluon._train_data_preprocess(model_name, warm_start=False)[source]#

Processing the data from self.trainer.datamodule for training.

Parameters:
model_name:

The name of a selected model.

warm_start

Finetune models based on previous trained models.

Returns:
dict

A dictionary that has the following keys: X_train, y_train, X_val, y_val, X_test, y_test. Those with postfixes _train or _val will be passed to _train_single_model() and _bayes_eval(). All of them will be passed to _pred_single_model() for evaluation.

Notes

self.trainer.datamodule.X_train/val/test are not scaled. To scale the df, run df = datamodule.data_transform(df, scaler_only=True)