[feat] Add coalescer

autoPyTorch/configs/greedy_portfolio.json

@@ -1,5 +1,6 @@
 [{"data_loader:batch_size": 60,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "NoFeaturePreprocessor",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -30,6 +31,7 @@
  "network_backbone:ShapedMLPBackbone:max_dropout": 0.023271935735825866},
  {"data_loader:batch_size": 255,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "NoFeaturePreprocessor",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -63,6 +65,7 @@
  "network_backbone:ShapedResNetBackbone:max_dropout": 0.7662454727603789},
  {"data_loader:batch_size": 165,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "NoFeaturePreprocessor",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -93,6 +96,7 @@
  "network_head:fully_connected:units_layer_1": 128},
  {"data_loader:batch_size": 299,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "NoFeaturePreprocessor",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -124,6 +128,7 @@
  "network_head:fully_connected:units_layer_1": 128},
  {"data_loader:batch_size": 183,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "NoFeaturePreprocessor",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -157,6 +162,7 @@
  "network_backbone:ShapedResNetBackbone:max_dropout": 0.27204101593048097},
  {"data_loader:batch_size": 21,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "NoFeaturePreprocessor",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -185,6 +191,7 @@
  "network_head:fully_connected:units_layer_1": 128},
  {"data_loader:batch_size": 159,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "TruncatedSVD",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -214,6 +221,7 @@
  "network_head:fully_connected:units_layer_1": 128},
  {"data_loader:batch_size": 442,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "TruncatedSVD",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -246,6 +254,7 @@
  "network_head:fully_connected:units_layer_1": 128},
  {"data_loader:batch_size": 140,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "TruncatedSVD",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -278,6 +287,7 @@
  "network_head:fully_connected:units_layer_1": 128},
  {"data_loader:batch_size": 48,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "NoFeaturePreprocessor",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -305,6 +315,7 @@
  "network_head:fully_connected:units_layer_1": 128},
  {"data_loader:batch_size": 168,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "NoFeaturePreprocessor",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -337,6 +348,7 @@
  "network_backbone:ShapedResNetBackbone:max_dropout": 0.8992826006547855},
  {"data_loader:batch_size": 21,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "NoFeaturePreprocessor",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -365,6 +377,7 @@
  "network_head:fully_connected:units_layer_1": 128},
  {"data_loader:batch_size": 163,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "NoFeaturePreprocessor",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -397,6 +410,7 @@
  "network_backbone:ShapedResNetBackbone:max_dropout": 0.6341848343636569},
  {"data_loader:batch_size": 150,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "NoFeaturePreprocessor",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -430,6 +444,7 @@
  "network_backbone:ShapedResNetBackbone:max_dropout": 0.7133813761319248},
  {"data_loader:batch_size": 151,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "TruncatedSVD",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",
@@ -459,6 +474,7 @@
  "network_head:fully_connected:units_layer_1": 128},
  {"data_loader:batch_size": 42,
  "encoder:__choice__": "OneHotEncoder",
+ "coalescer:__choice__": "NoCoalescer",
  "feature_preprocessor:__choice__": "TruncatedSVD",
  "imputer:numerical_strategy": "mean",
  "lr_scheduler:__choice__": "CosineAnnealingLR",

autoPyTorch/pipeline/components/preprocessing/tabular_preprocessing/coalescer/MinorityCoalescer.py

@@ -0,0 +1,44 @@
+from typing import Any, Dict, Optional, Union
+
+from ConfigSpace.configuration_space import ConfigurationSpace
+from ConfigSpace.hyperparameters import UniformFloatHyperparameter
+
+import numpy as np
+
+from autoPyTorch.pipeline.components.preprocessing.tabular_preprocessing.coalescer.base_coalescer import BaseCoalescer
+from autoPyTorch.utils.common import HyperparameterSearchSpace, add_hyperparameter
+from autoPyTorch.utils.implementations import MinorityCoalesceTransformer
+
+
+class MinorityCoalescer(BaseCoalescer):
+    """Group together categories whose occurence is less than a specified min_frac """
+    def __init__(self, min_frac: float, random_state: np.random.RandomState):
+        super().__init__()
+        self.min_frac = min_frac
+        self.random_state = random_state
+
+    def fit(self, X: Dict[str, Any], y: Any = None) -> BaseCoalescer:
+        self.check_requirements(X, y)
+        self.preprocessor['categorical'] = MinorityCoalesceTransformer(min_frac=self.min_frac)
+        return self
+
+    @staticmethod
+    def get_hyperparameter_search_space(
+        dataset_properties: Optional[Dict[str, Any]] = None,
+        min_frac: HyperparameterSearchSpace = HyperparameterSearchSpace(hyperparameter='min_frac',
+                                                                        value_range=(1e-4, 0.5),
+                                                                        default_value=1e-2,
+                                                                        ),
+    ) -> ConfigurationSpace:
+
+        cs = ConfigurationSpace()
+        add_hyperparameter(cs, min_frac, UniformFloatHyperparameter)
+        return cs
+
+    @staticmethod
+    def get_properties(dataset_properties: Optional[Dict[str, Any]] = None) -> Dict[str, Union[str, bool]]:
+        return {
+            'shortname': 'MinorityCoalescer',
+            'name': 'MinorityCoalescer',
+            'handles_sparse': False
+        }

autoPyTorch/pipeline/components/preprocessing/tabular_preprocessing/coalescer/NoCoalescer.py

@@ -0,0 +1,37 @@
+from typing import Any, Dict, Optional, Union
+
+import numpy as np
+
+from autoPyTorch.pipeline.components.preprocessing.tabular_preprocessing.coalescer.base_coalescer import BaseCoalescer
+
+
+class NoCoalescer(BaseCoalescer):
+    def __init__(self, random_state: np.random.RandomState):
+        super().__init__()
+        self.random_state = random_state
+        self._processing = False
+
+    def fit(self, X: Dict[str, Any], y: Optional[Any] = None) -> BaseCoalescer:
+        """
+        As no coalescing happens, only check the requirements.
+
+        Args:
+            X (Dict[str, Any]):
+                fit dictionary
+            y (Optional[Any]):
+                Parameter to comply with scikit-learn API. Not used.
+
+        Returns:
+            instance of self
+        """
+        self.check_requirements(X, y)
+
+        return self
+
+    @staticmethod
+    def get_properties(dataset_properties: Optional[Dict[str, Any]] = None) -> Dict[str, Union[str, bool]]:
+        return {
+            'shortname': 'NoCoalescer',
+            'name': 'NoCoalescer',
+            'handles_sparse': True
+        }