Add default params_to_tune for LabelEncoderTransform

CHANGELOG.md

@@ -55,6 +55,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Add default `params_to_tune` for `MedianOutliersTransform`, `DensityOutliersTransform` and `PredictionIntervalOutliersTransform` ([#1231](https://github.com/tinkoff-ai/etna/pull/1231))
 - Add default `params_to_tune` for `TimeSeriesImputerTransform` ([#1232](https://github.com/tinkoff-ai/etna/pull/1232))
 - Add default `params_to_tune` for `DifferencingTransform`, `MedianTransform`, `MaxTransform`, `MinTransform`, `QuantileTransform`, `StdTransform`, `MeanTransform`, `MADTransform`, `MinMaxDifferenceTransform`, `SumTransform`, `BoxCoxTransform`, `YeoJohnsonTransform`, `MaxAbsScalerTransform`, `MinMaxScalerTransform`, `RobustScalerTransform` and `StandardScalerTransform` ([#1233](https://github.com/tinkoff-ai/etna/pull/1233))
+- Add default `params_to_tune` for `LabelEncoderTransform` ([#1242](https://github.com/tinkoff-ai/etna/pull/1242))
 ### Fixed
 - Fix bug in `GaleShapleyFeatureSelectionTransform` with wrong number of remaining features ([#1110](https://github.com/tinkoff-ai/etna/pull/1110))
 - `ProphetModel` fails with additional seasonality set ([#1157](https://github.com/tinkoff-ai/etna/pull/1157))
@@ -68,6 +69,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Fix `DifferencingTransform` to raise error on new segments during `transform` and `inverse_transform` in inplace mode ([#1141](https://github.com/tinkoff-ai/etna/pull/1141))
 - Teach `DifferencingTransform` to `inverse_transform` with NaNs ([#1155](https://github.com/tinkoff-ai/etna/pull/1155))
 - Teach `BaseMixin.set_params` to work with nested `list` and `tuple` ([#1201](https://github.com/tinkoff-ai/etna/pull/1201))
+- 
 ### Removed
 - `sample_acf_plot`, `sample_pacf_plot`, `CatBoostModelPerSegment`, `CatBoostModelMultiSegment` ([#1118](https://github.com/tinkoff-ai/etna/pull/1118))
 - `PytorchForecastingTransform` ([#971](https://github.com/tinkoff-ai/etna/pull/971))

etna/transforms/encoders/categorical.py

@@ -1,4 +1,5 @@
 from enum import Enum
+from typing import Dict
 from typing import List
 from typing import Optional
 
@@ -8,9 +9,14 @@
 from sklearn.utils._encode import _check_unknown
 from sklearn.utils._encode import _encode
 
+from etna import SETTINGS
 from etna.datasets import TSDataset
 from etna.transforms.base import IrreversibleTransform
 
+if SETTINGS.auto_required:
+    from optuna.distributions import BaseDistribution
+    from optuna.distributions import CategoricalDistribution
+
 
 class ImputerMode(str, Enum):
     """Enum for different imputation strategy."""
@@ -130,6 +136,20 @@ def _get_column_name(self) -> str:
             return self.out_column
         return self.__repr__()
 
+    def params_to_tune(self) -> Dict[str, "BaseDistribution"]:
+        """Get default grid for tuning hyperparameters.
+
+        This grid tunes only ``strategy`` parameter. Other parameters are expected to be set by the user.
+
+        Returns
+        -------
+        :
+            Grid to tune.
+        """
+        return {
+            "strategy": CategoricalDistribution(["new_value", "mean"]),
+        }
+
 
 class OneHotEncoderTransform(IrreversibleTransform):
     """Encode categorical feature as a one-hot numeric features.

tests/test_transforms/test_encoders/test_categorical_transform.py

@@ -13,6 +13,7 @@
 from etna.transforms import FilterFeaturesTransform
 from etna.transforms.encoders.categorical import LabelEncoderTransform
 from etna.transforms.encoders.categorical import OneHotEncoderTransform
+from tests.test_transforms.utils import assert_sampling_is_valid
 from tests.test_transforms.utils import assert_transformation_equals_loaded_original
 
 
@@ -350,3 +351,16 @@ def test_save_load_ohe(dtype):
 def test_get_regressors_info_not_fitted(transform):
     with pytest.raises(ValueError, match="Fit the transform to get the correct regressors info!"):
         _ = transform.get_regressors_info()
+
+
+def test_params_to_tune_ohe():
+    transform = OneHotEncoderTransform(in_column="regressor_0")
+    assert len(transform.params_to_tune()) == 0
+
+
+def test_params_to_tune_label_encoder(ts_for_label_encoding):
+    ts, _ = ts_for_label_encoding
+    for i in range(3):
+        transform = LabelEncoderTransform(in_column=f"regressor_{i}", out_column="test")
+        assert len(transform.params_to_tune()) > 0
+        assert_sampling_is_valid(transform=transform, ts=ts)

tests/test_transforms/test_encoders/test_mean_segment_encoder_transform.py

@@ -98,3 +98,8 @@ def test_fit_transform_with_nans(ts_diff_endings):
 def test_save_load(almost_constant_ts):
     transform = MeanSegmentEncoderTransform()
     assert_transformation_equals_loaded_original(transform=transform, ts=almost_constant_ts)
+
+
+def test_params_to_tune():
+    transform = MeanSegmentEncoderTransform()
+    assert len(transform.params_to_tune()) == 0

tests/test_transforms/test_encoders/test_segment_encoder_transform.py

@@ -62,3 +62,8 @@ def test_new_segments_error(simple_ts):
 def test_save_load(example_tsds):
     transform = SegmentEncoderTransform()
     assert_transformation_equals_loaded_original(transform=transform, ts=example_tsds)
+
+
+def test_params_to_tune():
+    transform = SegmentEncoderTransform()
+    assert len(transform.params_to_tune()) == 0