Fix SegmentEncoderTransform to pass inference tests (#1103)

CHANGELOG.md

@@ -28,7 +28,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 -
 -
 -
--
+- Fix `SegmentEncoderTransform` to work with subset of segments and raise error on new segments ([#1103](https://github.com/tinkoff-ai/etna/pull/1103))
 -
 -
 ## [1.14.0] - 2022-12-16

etna/transforms/encoders/segment_encoder.py

@@ -1,3 +1,6 @@
+import reprlib
+
+import numpy as np
 import pandas as pd
 from sklearn import preprocessing
 
@@ -44,12 +47,31 @@ def transform(self, df: pd.DataFrame) -> pd.DataFrame:
         -------
         :
             result dataframe
+
+        Raises
+        ------
+        ValueError:
+            If transform isn't fitted.
+        ValueError:
+            If there are segments that weren't present during training.
         """
-        encoded_matrix = self._le.transform(self._le.classes_)
-        encoded_matrix = encoded_matrix.reshape(len(self._le.classes_), -1).repeat(len(df), axis=1).T
+        segments = df.columns.get_level_values("segment").unique().tolist()
+
+        try:
+            new_segments = set(segments) - set(self._le.classes_)
+        except AttributeError:
+            raise ValueError("The transform isn't fitted!")
+
+        if len(new_segments) > 0:
+            raise ValueError(
+                f"This transform can't process segments that weren't present on train data: {reprlib.repr(new_segments)}"
+            )
+
+        encoded_matrix = self._le.transform(segments)
+        encoded_matrix = np.tile(encoded_matrix, (len(df), 1))
         encoded_df = pd.DataFrame(
             encoded_matrix,
-            columns=pd.MultiIndex.from_product([self._le.classes_, ["segment_code"]], names=("segment", "feature")),
+            columns=pd.MultiIndex.from_product([segments, ["segment_code"]], names=("segment", "feature")),
             index=df.index,
         )
         encoded_df = encoded_df.astype("category")

tests/test_transforms/test_encoders/test_segment_encoder_transform.py

@@ -1,5 +1,6 @@
 import numpy as np
 import pandas as pd
+import pytest
 
 from etna.transforms import SegmentEncoderTransform
 from tests.test_transforms.utils import assert_transformation_equals_loaded_original
@@ -21,6 +22,38 @@ def test_segment_encoder_transform(dummy_df):
     assert codes == {0, 1}, "Codes are not 0 and 1"
 
 
+def test_subset_segments(dummy_df):
+    train_df = dummy_df
+    test_df = dummy_df.loc[:, pd.IndexSlice["Omsk", :]]
+    transform = SegmentEncoderTransform()
+
+    transform.fit(train_df)
+    transformed_test_df = transform.transform(test_df)
+
+    segments = sorted(transformed_test_df.columns.get_level_values("segment").unique())
+    features = sorted(transformed_test_df.columns.get_level_values("feature").unique())
+    assert segments == ["Omsk"]
+    assert features == ["segment_code", "target"]
+    values = transformed_test_df.loc[:, pd.IndexSlice[:, "segment_code"]]
+    assert np.all(values == values.iloc[0])
+
+
+def test_not_fitted_error(dummy_df):
+    encoder = SegmentEncoderTransform()
+    with pytest.raises(ValueError, match="The transform isn't fitted"):
+        encoder.transform(dummy_df)
+
+
+def test_new_segments_error(dummy_df):
+    train_df = dummy_df.loc[:, pd.IndexSlice["Moscow", :]]
+    test_df = dummy_df.loc[:, pd.IndexSlice["Omsk", :]]
+    transform = SegmentEncoderTransform()
+
+    transform.fit(train_df)
+    with pytest.raises(ValueError, match="This transform can't process segments that weren't present on train data"):
+        _ = transform.transform(test_df)
+
+
 def test_save_load(example_tsds):
     transform = SegmentEncoderTransform()
     assert_transformation_equals_loaded_original(transform=transform, ts=example_tsds)