Change the way SARIMAXModel works with regressors (#380)

* Update _SARIMAX methods

* Update tests

etna/models/sarimax.py

@@ -165,21 +165,24 @@ def __init__(
         self.kwargs = kwargs
         self._model: Optional[SARIMAX] = None
         self._result: Optional[SARIMAX] = None
+        self.regressor_columns: Optional[List[str]] = None
 
-    def fit(self, df: pd.DataFrame) -> "_SARIMAXModel":
+    def fit(self, df: pd.DataFrame, regressors: Optional[List[str]]) -> "_SARIMAXModel":
         """
         Fits a SARIMAX model.
 
         Parameters
         ----------
         df:
             Features dataframe
-
+        regressors:
+            List of the columns with regressors
         Returns
         -------
         self: SARIMAX
             fitted model
         """
+        self.regressor_columns = regressors
         categorical_cols = df.select_dtypes(include=["category"]).columns.tolist()
         try:
             df.loc[:, categorical_cols] = df[categorical_cols].astype(int)
@@ -278,27 +281,23 @@ def predict(self, df: pd.DataFrame, prediction_interval: bool, quantiles: Sequen
         return y_pred.reset_index(drop=True, inplace=False)
 
     def _check_df(self, df: pd.DataFrame, horizon: Optional[int] = None):
-        column_to_drop = [
-            col for col in df.columns if not col.startswith("regressor") and col not in ["target", "timestamp"]
-        ]
-        regressor_columns = [col for col in df.columns if col.startswith("regressor")]
+        column_to_drop = [col for col in df.columns if col not in ["target", "timestamp"] + self.regressor_columns]
         if column_to_drop:
             warnings.warn(
                 message=f"SARIMAX model does not work with exogenous features (features unknown in future).\n "
                 f"{column_to_drop} will be dropped"
             )
         if horizon:
-            short_regressors = [regressor for regressor in regressor_columns if df[regressor].count() < horizon]
+            short_regressors = [regressor for regressor in self.regressor_columns if df[regressor].count() < horizon]
             if short_regressors:
                 raise ValueError(
                     f"Regressors {short_regressors} are too short for chosen horizon value.\n "
                     "Try lower horizon value, or drop this regressors."
                 )
 
     def _select_regressors(self, df: pd.DataFrame) -> Optional[pd.DataFrame]:
-        regressor_columns = [col for col in df.columns if col.startswith("regressor")]
-        if regressor_columns:
-            exog_future = df[regressor_columns]
+        if self.regressor_columns:
+            exog_future = df[self.regressor_columns]
             exog_future.index = df["timestamp"]
         else:
             exog_future = None
@@ -469,6 +468,21 @@ def __init__(
             )
         )
 
+    @log_decorator
+    def fit(self, ts: TSDataset) -> "SARIMAXModel":
+        """Fit model."""
+        self._segments = ts.segments
+        self._build_models()
+
+        for segment in self._segments:
+            model = self._models[segment]
+            segment_features = ts[:, segment, :]
+            segment_features = segment_features.dropna()
+            segment_features = segment_features.droplevel("segment", axis=1)
+            segment_features = segment_features.reset_index()
+            model.fit(df=segment_features, regressors=ts.regressors)
+        return self
+
     @staticmethod
     def _forecast_one_segment(
         model,

tests/test_models/test_sarimax_model.py

@@ -21,6 +21,23 @@ def test_sarimax_forecaster_run(example_tsds):
     assert len(res) == 14
 
 
+def test_sarimax_save_regressors_on_fit(example_reg_tsds):
+    model = SARIMAXModel()
+    model.fit(ts=example_reg_tsds)
+    for segment_model in model._models.values():
+        assert sorted(segment_model.regressor_columns) == example_reg_tsds.regressors
+
+
+def test_sarimax_select_regressors_correctly(example_reg_tsds):
+    model = SARIMAXModel()
+    model.fit(ts=example_reg_tsds)
+    for segment, segment_model in model._models.items():
+        segment_features = example_reg_tsds[:, segment, :].droplevel("segment", axis=1)
+        segment_regressors_expected = segment_features[example_reg_tsds.regressors]
+        segment_regressors = segment_model._select_regressors(df=segment_features.reset_index())
+        assert (segment_regressors == segment_regressors_expected).all().all()
+
+
 def test_sarimax_forecaster_run_with_reg(example_reg_tsds):
     """
     Given: I have dataframe with 2 segments