List of tabular models update (#1379)

* work: added deprecated tag

* work: set 'boosting' tag for gbm

* docs: available tabular models

* small fix

* work: tests adaptation, closed old todo

* fix: test update

* work: update specific operators

* fix: test models repository update

* chore: operation repository update

* Revert "chore: operation repository update"

This reverts commit e81c8232.

* chore: test_result_reproducing params update

* chore: test_presets update

* docs: available models update

* fix: test_composer update

* docs: decription update

* docs: available models update

* fix: dask-ml for python3.8 version update

* fix: 2023.3.24 dask-ml version for py3.8

* Update requirements.txt

* Update requirements.txt

* Update requirements.txt

* fix: fsspec==2023.1.0

* fix: update dask, dask-ml reqs

* smallfix

* Update requirements.txt

fedot/api/builder.py

@@ -258,7 +258,7 @@ class FedotBuilder:
         Args:
             available_operations: list of model names to use. Pick the names according to operations repository.
 
-                .. details:: Possible options:
+                .. details:: All options:
 
                     - ``adareg`` -> AdaBoost Regressor
                     - ``ar`` -> AutoRegression
@@ -331,6 +331,25 @@ class FedotBuilder:
                     - ``exog_ts`` -> Exogeneus Transformation
                     - ``topological_features`` -> Topological features
 
+                .. details:: Tabular models available for composing:
+
+                    - ``adareg`` -> AdaBoost Regressor
+                    - ``catboost`` -> Catboost Classifier
+                    - ``catboostreg`` -> Catboost Regressor
+                    - ``knn`` -> K-nearest neighbors Classifier
+                    - ``knnreg`` -> K-nearest neighbors Regressor
+                    - ``lasso`` -> Lasso Linear Regressor
+                    - ``lgbm`` -> Light Gradient Boosting Machine Classifier
+                    - ``lgbmreg`` -> Light Gradient Boosting Machine Regressor
+                    - ``linear`` -> Linear Regression Regressor
+                    - ``logit`` -> Logistic Regression Classifier
+                    - ``rf`` -> Random Forest Classifier
+                    - ``rfr`` -> Random Forest Regressor
+                    - ``ridge`` -> Ridge Linear Regressor
+                    - ``treg`` -> Extra Trees Regressor
+                    - ``xgboost`` -> Extreme Gradient Boosting Classifier
+                    - ``xgbreg`` -> Extreme Gradient Boosting Regressor
+
             max_depth: max depth of a pipeline. Defaults to ``6``.
 
             max_arity: max arity of a pipeline nodes. Defaults to ``3``.

fedot/core/composer/gp_composer/specific_operators.py

@@ -111,13 +111,13 @@ def add_resample_mutation(pipeline: Pipeline, **kwargs):
 
 
 def choose_new_model(boosting_model_candidates: List[str]) -> str:
-    """ Since 'linear' and 'dtreg' operations are suitable for solving the problem
+    """ Since 'ridge' and 'rfr' operations are suitable for solving the problem
     and they are simpler than others, they are preferred """
 
-    if 'linear' in boosting_model_candidates:
-        new_model = 'linear'
-    elif 'dtreg' in boosting_model_candidates:
-        new_model = 'dtreg'
+    if 'ridge' in boosting_model_candidates:
+        new_model = 'ridge'
+    elif 'rfr' in boosting_model_candidates:
+        new_model = 'rfr'
     else:
         new_model = choice(boosting_model_candidates)
     return new_model

fedot/core/repository/data/model_repository.json

@@ -211,6 +211,7 @@
       "meta": "sklearn_class",
       "presets": ["fast_train"],
       "tags": [
+        "deprecated",
         "simple",
         "bayesian",
         "non_multi",
@@ -235,6 +236,7 @@
       "meta": "sklearn_class",
       "presets": ["fast_train", "*tree"],
       "tags": [
+        "deprecated",
         "simple",
         "tree",
         "interpretable",
@@ -245,6 +247,7 @@
       "meta": "sklearn_regr",
       "presets": ["fast_train", "ts", "*tree"],
       "tags": [
+        "deprecated",
         "tree",
         "interpretable",
         "non_linear"
@@ -254,6 +257,7 @@
       "meta": "sklearn_regr",
       "presets": ["*tree"],
       "tags": [
+        "deprecated",
         "boosting",
         "non_multi",
         "non_linear",
@@ -296,20 +300,20 @@
       "meta": "custom_class",
       "presets": ["fast_train"],
       "tags": [
-        "discriminant", "linear", "correct_params", "non-default"
+        "deprecated", "discriminant", "linear", "correct_params", "non-default"
       ]
     },
     "lgbm": {
       "meta": "boosting_class",
       "tags": [
-        "tree", "non_linear"
+        "tree", "non_linear", "boosting"
       ]
     },
     "lgbmreg": {
       "meta": "boosting_regr",
       "presets": ["*tree"],
       "tags": [
-        "tree", "non_multi", "non_linear"
+        "tree", "non_multi", "non_linear", "boosting"
       ]
     },
     "linear": {
@@ -332,6 +336,7 @@
     "mlp": {
       "meta": "sklearn_class",
       "tags": [
+        "deprecated",
         "neural",
         "non_linear"
       ]
@@ -340,6 +345,7 @@
       "meta": "sklearn_class",
       "presets": ["fast_train"],
       "tags": [
+        "deprecated",
         "non-default",
         "bayesian",
         "non_multi",
@@ -350,6 +356,7 @@
       "meta": "custom_class",
       "presets": ["fast_train"],
       "tags": [
+        "deprecated",
         "discriminant",
         "quadratic",
         "non_linear"
@@ -390,7 +397,7 @@
       "meta": "sklearn_regr",
       "presets": ["fast_train", "ts"],
       "tags": [
-        "non_multi", "non_linear"
+        "deprecated", "non_multi", "non_linear"
       ]
     },
     "stl_arima": {
@@ -455,6 +462,7 @@
     "svc": {
       "meta": "custom_class",
       "tags": [
+        "deprecated",
         "no_prob",
         "expensive",
         "non_linear"
@@ -463,6 +471,7 @@
     "svr": {
       "meta": "sklearn_regr",
       "tags": [
+        "deprecated",
         "non_multi",
         "non_linear"
       ]
@@ -480,14 +489,14 @@
       "meta": "boosting_class",
       "presets": ["*tree"],
       "tags": [
-        "tree", "non_linear"
+        "tree", "non_linear", "boosting"
       ]
     },
     "xgboostreg": {
       "meta": "boosting_regr",
       "presets": ["*tree"],
       "tags": [
-        "tree", "non_multi", "non_linear"
+        "tree", "non_multi", "non_linear", "boosting"
       ]
     },
     "cnn": {

fedot/core/repository/operation_types_repository.py

@@ -89,7 +89,7 @@ class OperationTypesRepository:
     def __init__(self, operation_type: str = 'model'):
         self.log = default_log(self)
 
-        self._tags_excluded_by_default = ['non-default', 'expensive']
+        self._tags_excluded_by_default = ['non-default', 'expensive', 'deprecated']
         OperationTypesRepository.init_default_repositories()
 
         self.operation_type = operation_type

requirements.txt

@@ -7,8 +7,9 @@ thegolem==0.4.1
 # Data
 numpy>=1.16.0, !=1.24.0
 pandas>=1.3.0; python_version >='3.8'
-dask-ml>=2024.4.4; python_version >= '3.10'
-dask-ml>=2023.5.0; python_version < '3.10'
+dask-ml==2024.4.4; python_version <= '3.10'
+dask-ml>2024.4.4; python_version > '3.10'
+dask>=2023.5.0; python_version < '3.10'
 
 # Models and frameworks
 anytree>=2.8.0
@@ -38,6 +39,8 @@ joblib>=0.17.0
 requests>=2.0
 typing>=3.7.0
 psutil>=5.9.2
+fsspec>=2024; python_version > '3.8'
+fsspec>=2024,<=2025.3.0; python_version <= '3.8'
 
 # Tests
 pytest>=6.2.0

test/integration/api/test_api_utils.py

@@ -84,13 +84,13 @@ def test_the_formation_of_initial_assumption():
 
     train_input, _, _ = get_dataset(task_type='classification')
     train_input = DataPreprocessor().obligatory_prepare_for_fit(train_input)
-    available_operations = ['dt']
+    available_operations = ['rf']
 
     initial_assumptions = AssumptionsBuilder \
         .get(train_input) \
         .from_operations(available_operations) \
         .build()
-    res_init_assumption = Pipeline(PipelineNode('dt'))
+    res_init_assumption = Pipeline(PipelineNode('rf'))
     assert initial_assumptions[0].root_node.descriptive_id == res_init_assumption.root_node.descriptive_id
 
 

test/integration/api/test_main_api.py

@@ -89,8 +89,8 @@ def test_api_tune_correct(task_type, metric_name, pred_model):
 @pytest.mark.parametrize(
     "task_type, metric_name, pred_model",
     [
-        ("classification", "f1", "dt"),
-        ("regression", "rmse", "dtreg"),
+        ("classification", "f1", "rf"),
+        ("regression", "rmse", "rfr"),
     ],
 )
 def test_api_fit_atomized_model(task_type, metric_name, pred_model):

test/integration/real_applications/test_model_result_reproducing.py

@@ -30,8 +30,8 @@ def get_fitted_fedot(forecast_length, train_data, **kwargs):
               'task_params': TsForecastingParams(forecast_length=forecast_length),
               'seed': 1,
               'timeout': None,
-              'pop_size': 5,
-              'num_of_generations': 5,
+              'pop_size': 4,
+              'num_of_generations': 4,
               'with_tuning': False}
     params.update(kwargs)
     fedot = Fedot(**params)

test/unit/api/test_api_safety.py

@@ -100,7 +100,7 @@ def test_api_fit_predict_with_pseudo_large_dataset_with_label_correct():
     model.predict(features=data)
 
     # there should be only tree like models + data operations
-    assert len(model.params.get('available_operations')) == 5
+    assert len(model.params.get('available_operations')) == 4
     assert 'logit' not in model.params.get('available_operations')
 
 

test/unit/api/test_presets.py

@@ -12,8 +12,8 @@ from test.data.datasets import data_with_binary_features_and_categorical_target
 
 
 minimal_sets_for_fast_train_by_task = {
-    'classification': {'dt', 'logit', 'knn'},
-    'regression': {'dtreg', 'lasso', 'ridge', 'linear'},
+    'classification': {'rf', 'logit', 'knn'},
+    'regression': {'rfr', 'lasso', 'ridge', 'linear'},
     'ts_forecasting': {'ar', 'adareg', 'scaling', 'lasso'},
     'clustering': {'kmeans'}
 }

test/unit/optimizer/gp_operators/test_mutation.py

@@ -74,7 +74,7 @@ def get_simple_linear_boosting_pipeline() -> Pipeline:
     node_pf = PipelineNode('poly_features', nodes_from=[node_scaling])
     node_rf = PipelineNode('rf', nodes_from=[node_pf])
     node_decompose = PipelineNode('class_decompose', nodes_from=[node_pf, node_rf])
-    node_linear = PipelineNode('linear', nodes_from=[node_decompose])
+    node_linear = PipelineNode('ridge', nodes_from=[node_decompose])
     final_node = PipelineNode('logit', nodes_from=[node_linear, node_rf])
     pipeline = Pipeline(final_node)
     return pipeline