Louquinze development (#1578)

* create new text preprocessing cs

* create new text preprocessing cs

* set new defaults for text encoding

* set new defaults for text encoding

* set new defaults for text encoding

* Fix bug, rework tests

Co-authored-by: lukas <[email protected]>

Author: mfeurer

autosklearn/pipeline/components/data_preprocessing/feature_type.py

@@ -267,7 +267,9 @@ def set_hyperparameters(
                     sub_config_dict[new_name] = value
 
             sub_configuration = Configuration(
-                sub_configuration_space, values=sub_config_dict
+                sub_configuration_space,
+                values=sub_config_dict,
+                allow_inactive_with_values=True,
             )
 
             sub_init_params_dict: Optional[Dict[str, Any]] = None

autosklearn/pipeline/components/data_preprocessing/feature_type_categorical.py

@@ -33,12 +33,10 @@ class CategoricalPreprocessingPipeline(BasePipeline):
         3 - Minority coalescence: Assign category 1 to all categories whose occurrence
             don't sum-up to a certain minimum fraction
         4 - One hot encoding: usual sklearn one hot encoding
-
     Parameters
     ----------
     config : ConfigSpace.configuration_space.Configuration
         The configuration to evaluate.
-
     random_state : Optional[int | RandomState]
         If int, random_state is the seed used by the random number generator;
         If RandomState instance, random_state is the random number generator;
@@ -101,7 +99,6 @@ def _get_hyperparameter_search_space(
         dataset_properties: Optional[DATASET_PROPERTIES_TYPE] = None,
     ) -> ConfigurationSpace:
         """Create the hyperparameter configuration space.
-
         Returns
         -------
         cs : ConfigSpace.configuration_space.Configuration

autosklearn/pipeline/components/data_preprocessing/feature_type_numerical.py

@@ -26,12 +26,10 @@ class NumericalPreprocessingPipeline(BasePipeline):
         2 - VarianceThreshold: Removes low-variance features
         3 - Rescaling: rescale features according to a certain rule (e.g. normalization,
             standartization or min-max)
-
     Parameters
     ----------
     config : ConfigSpace.configuration_space.Configuration
         The configuration to evaluate.
-
     random_state : Optional[int | RandomState]
         If int, random_state is the seed used by the random number generator;
         If RandomState instance, random_state is the random number generator;
@@ -94,10 +92,8 @@ def _get_hyperparameter_search_space(
         dataset_properties: Optional[DATASET_PROPERTIES_TYPE] = None,
     ) -> ConfigurationSpace:
         """Create the hyperparameter configuration space.
-
         Parameters
         ----------
-
         Returns
         -------
         cs : ConfigSpace.configuration_space.Configuration

autosklearn/pipeline/components/data_preprocessing/text_encoding/__init__.py

@@ -63,7 +63,7 @@ def get_hyperparameter_search_space(
             )
 
         if default is None:
-            defaults = ["bag_of_words_encoding"]
+            defaults = ["tfidf_encoding"]
             for default_ in defaults:
                 if default_ in available_preprocessors:
                     default = default_