require passing in date_format for date_columns

src/pytorch_tabular/config/config.py

@@ -65,8 +65,9 @@ class DataConfig:
         categorical_cols (List): Column names of the categorical fields to treat differently. Defaults to
                 []
 
-        date_columns (List): (Column names, Freq) tuples of the date fields. For eg. a field named
-                introduction_date and with a monthly frequency should have an entry ('intro_date','M'}
+        date_columns (List): (Column name, Freq, Format) tuples of the date fields. For eg. a field named
+                introduction_date and with a monthly frequency like "2023-12" should have
+                an entry ('intro_date','M','%Y-%m')
 
         encode_date_columns (bool): Whether or not to encode the derived variables from date
 
@@ -115,7 +116,8 @@ class DataConfig:
         default_factory=list,
         metadata={
             "help": "(Column names, Freq) tuples of the date fields. For eg. a field named"
-            " `introduction_date` and with a monthly frequency should have an entry ('intro_date','M'}"
+            " introduction_date and with a monthly frequency like '2023-12' should have"
+            " an entry ('intro_date','M','%Y-%m')"
         },
     )
 

src/pytorch_tabular/tabular_datamodule.py

@@ -287,8 +287,8 @@ def do_leave_one_out_encoder(self) -> bool:
 
     def _encode_date_columns(self, data: DataFrame) -> DataFrame:
         added_features = []
-        for field_name, freq in self.config.date_columns:
-            data = self.make_date(data, field_name)
+        for field_name, freq, format in self.config.date_columns:
+            data = self.make_date(data, field_name, format)
             data, _new_feats = self.add_datepart(data, field_name, frequency=freq, prefix=None, drop=True)
             added_features += _new_feats
         return data, added_features
@@ -630,7 +630,7 @@ def time_features_from_frequency_str(cls, freq_str: str) -> List[str]:
 
     # adapted from fastai
     @classmethod
-    def make_date(cls, df: DataFrame, date_field: str) -> DataFrame:
+    def make_date(cls, df: DataFrame, date_field: str, date_format: str = "ISO8601") -> DataFrame:
         """Make sure `df[date_field]` is of the right date type.
 
         Args:
@@ -645,7 +645,7 @@ def make_date(cls, df: DataFrame, date_field: str) -> DataFrame:
         if isinstance(field_dtype, DatetimeTZDtype):
             field_dtype = np.datetime64
         if not np.issubdtype(field_dtype, np.datetime64):
-            df[date_field] = to_datetime(df[date_field], infer_datetime_format=True)
+            df[date_field] = to_datetime(df[date_field], format=date_format)
         return df
 
     # adapted from fastai

tests/test_datamodule.py

@@ -117,7 +117,7 @@ def test_date_encoding(timeseries_data, freq):
         target=target + ["Occupancy"],
         continuous_cols=["Temperature", "Humidity", "Light", "CO2", "HumidityRatio"],
         categorical_cols=[],
-        date_columns=[("date", freq)],
+        date_columns=[("date", freq, "%Y-%m-%d %H:%M:%S")],
         encode_date_columns=True,
     )
     model_config_params = {"task": "regression"}