Bug fix for "Categorical" dtype (#493)

* Categorical bug fix

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* dataloader kwargs part removed

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Manu Joseph V <manujosephv@gmail.com>

Author: 3 people

src/pytorch_tabular/categorical_encoders.py

@@ -62,6 +62,8 @@ def transform(self, X):
                 not X[self.cols].isnull().any().any()
             ), "`handle_missing` = `error` and missing values found in columns to encode."
         X_encoded = X.copy(deep=True)
+        category_cols = X_encoded.select_dtypes(include="category").columns
+        X_encoded[category_cols] = X_encoded[category_cols].astype("object")
         for col, mapping in self._mapping.items():
             X_encoded[col] = X_encoded[col].fillna(NAN_CATEGORY).map(mapping["value"])
 

src/pytorch_tabular/tabular_datamodule.py

@@ -301,10 +301,14 @@ def _update_config(self, config) -> InferredConfig:
         else:
             raise ValueError(f"{config.task} is an unsupported task.")
         if self.train is not None:
+            category_cols = self.train[config.categorical_cols].select_dtypes(include="category").columns
+            self.train[category_cols] = self.train[category_cols].astype("object")
             categorical_cardinality = [
                 int(x) + 1 for x in list(self.train[config.categorical_cols].fillna("NA").nunique().values)
             ]
         else:
+            category_cols = self.train_dataset.data[config.categorical_cols].select_dtypes(include="category").columns
+            self.train_dataset.data[category_cols] = self.train_dataset.data[category_cols].astype("object")
             categorical_cardinality = [
                 int(x) + 1 for x in list(self.train_dataset.data[config.categorical_cols].nunique().values)
             ]