autogluon
autogluon copied to clipboard
[ENHANCEMENT] TimeSeriesPredictor doesnot support "ignored_columns" similar to Tabular Predictor
Tabluar predictor takes ignored_columns as input, but TimeSeriesPredictor doesnot. It will be helpful to have support for TimeSeriesPredictor too
I think the easier option is just to drop them from the training data when calling fit, not sure if we need to extend the API for this
predictor.fit(train_data.drop(ignored_columns, axis=1))
I will be nice to have those ignored columns in train data and not drop them. Tabular predictor support its. Api should be uniform across different predictors if possible