cross_validation.py has problems when ran as main.

[matsantoro]

In cross_validation.py, CrossValidate.cross_validate() has comparison between strings and pd.Timestamp (lines 140-147) if called as main. One can add

df['date'] = pd.to_datetime(df['date'])

on line 234.

The function extract_features_for_prediction doesn't work when cross_validation.py is ran as main, because it attempts to stack the train dataset with topological features to the test dataset. Substitute line 36 including only the non-topological features:

all_x_train = x_train[:, :14]

Finally, on line 200, one should have

model_config_with_topo = self._validate_k_fold_model(x_train_model[:train_x.shape[0]], train_y, val_x_with_topo, val_y)

as x_train_model is the stacking of the original x_train with the validation features.