Model builders API update (#1320)

---------

Co-authored-by: Dmitry Razdoburdin <>
Co-authored-by: Nikolay Petrov <nikolay.a.petrov@intel.com>
Co-authored-by: Alexander Andreev <alexander.andreev@intel.com>

Author: 3 people

daal4py/__init__.py

@@ -49,3 +49,8 @@
             '[c]sh/psxevars.[c]sh may solve the issue.\n')
 
     raise
+
+from . import mb
+from . import sklearn
+
+__all__ = ['mb', 'sklearn']

daal4py/mb/__init__.py

@@ -0,0 +1,20 @@
+#!/usr/bin/env python
+#===============================================================================
+# Copyright 2023 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#===============================================================================
+
+from .model_builders import GBTDAALBaseModel, convert_model
+
+__all__ = ['GBTDAALBaseModel', 'convert_model']

daal4py/mb/model_builders.py

@@ -0,0 +1,222 @@
+#===============================================================================
+# Copyright 2023 Intel Corporation
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#===============================================================================
+
+# daal4py Model builders API
+
+import numpy as np
+import daal4py as d4p
+
+try:
+    from pandas import DataFrame
+    from pandas.core.dtypes.cast import find_common_type
+    pandas_is_imported = True
+except (ImportError, ModuleNotFoundError):
+    pandas_is_imported = False
+
+
+def parse_dtype(dt):
+    if dt == np.double:
+        return "double"
+    if dt == np.single:
+        return "float"
+    raise ValueError(f"Input array has unexpected dtype = {dt}")
+
+
+def getFPType(X):
+    if pandas_is_imported:
+        if isinstance(X, DataFrame):
+            dt = find_common_type(X.dtypes.tolist())
+            return parse_dtype(dt)
+
+    dt = getattr(X, 'dtype', None)
+    return parse_dtype(dt)
+
+
+class GBTDAALBaseModel:
+    def _get_params_from_lightgbm(self, params):
+        self.n_classes_ = params["num_tree_per_iteration"]
+        objective_fun = params["objective"]
+        if self.n_classes_ <= 2:
+            if "binary" in objective_fun:  # nClasses == 1
+                self.n_classes_ = 2
+
+        self.n_features_in_ = params["max_feature_idx"] + 1
+
+    def _get_params_from_xgboost(self, params):
+        self.n_classes_ = int(params["learner"]["learner_model_param"]["num_class"])
+        objective_fun = params["learner"]["learner_train_param"]["objective"]
+        if self.n_classes_ <= 2:
+            if objective_fun in ["binary:logistic", "binary:logitraw"]:
+                self.n_classes_ = 2
+
+        self.n_features_in_ = int(params["learner"]["learner_model_param"]["num_feature"])
+
+    def _get_params_from_catboost(self, params):
+        if 'class_params' in params['model_info']:
+            self.n_classes_ = len(params['model_info']['class_params']['class_to_label'])
+        self.n_features_in_ = len(params['features_info']['float_features'])
+
+    def _convert_model_from_lightgbm(self, booster):
+        lgbm_params = d4p.get_lightgbm_params(booster)
+        self.daal_model_ = d4p.get_gbt_model_from_lightgbm(booster, lgbm_params)
+        self._get_params_from_lightgbm(lgbm_params)
+
+    def _convert_model_from_xgboost(self, booster):
+        xgb_params = d4p.get_xgboost_params(booster)
+        self.daal_model_ = d4p.get_gbt_model_from_xgboost(booster, xgb_params)
+        self._get_params_from_xgboost(xgb_params)
+
+    def _convert_model_from_catboost(self, booster):
+        catboost_params = d4p.get_catboost_params(booster)
+        self.daal_model_ = d4p.get_gbt_model_from_catboost(booster)
+        self._get_params_from_catboost(catboost_params)
+
+    def _convert_model(self, model):
+        (submodule_name, class_name) = (model.__class__.__module__,
+                                        model.__class__.__name__)
+        self_class_name = self.__class__.__name__
+
+        # Build GBTDAALClassifier from LightGBM
+        if (submodule_name, class_name) == ("lightgbm.sklearn", "LGBMClassifier"):
+            if self_class_name == "GBTDAALClassifier":
+                self._convert_model_from_lightgbm(model.booster_)
+            else:
+                raise TypeError(f"Only GBTDAALClassifier can be created from\
+                                 {submodule_name}.{class_name} (got {self_class_name})")
+        # Build GBTDAALClassifier from XGBoost
+        elif (submodule_name, class_name) == ("xgboost.sklearn", "XGBClassifier"):
+            if self_class_name == "GBTDAALClassifier":
+                self._convert_model_from_xgboost(model.get_booster())
+            else:
+                raise TypeError(f"Only GBTDAALClassifier can be created from\
+                                 {submodule_name}.{class_name} (got {self_class_name})")
+        # Build GBTDAALClassifier from CatBoost
+        elif (submodule_name, class_name) == ("catboost.core", "CatBoostClassifier"):
+            if self_class_name == "GBTDAALClassifier":
+                self._convert_model_from_catboost(model)
+            else:
+                raise TypeError(f"Only GBTDAALClassifier can be created from\
+                                 {submodule_name}.{class_name} (got {self_class_name})")
+        # Build GBTDAALRegressor from LightGBM
+        elif (submodule_name, class_name) == ("lightgbm.sklearn", "LGBMRegressor"):
+            if self_class_name == "GBTDAALRegressor":
+                self._convert_model_from_lightgbm(model.booster_)
+            else:
+                raise TypeError(f"Only GBTDAALRegressor can be created from\
+                                 {submodule_name}.{class_name} (got {self_class_name})")
+        # Build GBTDAALRegressor from XGBoost
+        elif (submodule_name, class_name) == ("xgboost.sklearn", "XGBRegressor"):
+            if self_class_name == "GBTDAALRegressor":
+                self._convert_model_from_xgboost(model.get_booster())
+            else:
+                raise TypeError(f"Only GBTDAALRegressor can be created from\
+                                 {submodule_name}.{class_name} (got {self_class_name})")
+        # Build GBTDAALRegressor from CatBoost
+        elif (submodule_name, class_name) == ("catboost.core", "CatBoostRegressor"):
+            if self_class_name == "GBTDAALRegressor":
+                self._convert_model_from_catboost(model)
+            else:
+                raise TypeError(f"Only GBTDAALRegressor can be created from\
+                                 {submodule_name}.{class_name} (got {self_class_name})")
+        # Build GBTDAALModel from LightGBM
+        elif (submodule_name, class_name) == ("lightgbm.basic", "Booster"):
+            if self_class_name == "GBTDAALModel":
+                self._convert_model_from_lightgbm(model)
+            else:
+                raise TypeError(f"Only GBTDAALModel can be created from\
+                                 {submodule_name}.{class_name} (got {self_class_name})")
+        # Build GBTDAALModel from XGBoost
+        elif (submodule_name, class_name) == ("xgboost.core", "Booster"):
+            if self_class_name == "GBTDAALModel":
+                self._convert_model_from_xgboost(model)
+            else:
+                raise TypeError(f"Only GBTDAALModel can be created from\
+                                 {submodule_name}.{class_name} (got {self_class_name})")
+        # Build GBTDAALModel from CatBoost
+        elif (submodule_name, class_name) == ("catboost.core", "CatBoost"):
+            if self_class_name == "GBTDAALModel":
+                self._convert_model_from_catboost(model)
+            else:
+                raise TypeError(f"Only GBTDAALModel can be created from\
+                                 {submodule_name}.{class_name} (got {self_class_name})")
+        else:
+            raise TypeError(f"Unknown model format {submodule_name}.{class_name}")
+
+    def _predict_classification(self, X, fptype, resultsToEvaluate):
+        if X.shape[1] != self.n_features_in_:
+            raise ValueError('Shape of input is different from what was seen in `fit`')
+
+        if not hasattr(self, 'daal_model_'):
+            raise ValueError((
+                "The class {} instance does not have 'daal_model_' attribute set. "
+                "Call 'fit' with appropriate arguments before using this method.")
+                .format(type(self).__name__))
+
+        # Prediction
+        predict_algo = d4p.gbt_classification_prediction(
+            fptype=fptype,
+            nClasses=self.n_classes_,
+            resultsToEvaluate=resultsToEvaluate)
+        predict_result = predict_algo.compute(X, self.daal_model_)
+
+        if resultsToEvaluate == "computeClassLabels":
+            return predict_result.prediction.ravel().astype(np.int64, copy=False)
+        else:
+            return predict_result.probabilities
+
+    def _predict_regression(self, X, fptype):
+        if X.shape[1] != self.n_features_in_:
+            raise ValueError('Shape of input is different from what was seen in `fit`')
+
+        if not hasattr(self, 'daal_model_'):
+            raise ValueError((
+                "The class {} instance does not have 'daal_model_' attribute set. "
+                "Call 'fit' with appropriate arguments before using this method.").format(
+                    type(self).__name__))
+
+        # Prediction
+        predict_algo = d4p.gbt_regression_prediction(fptype=fptype)
+        predict_result = predict_algo.compute(X, self.daal_model_)
+
+        return predict_result.prediction.ravel()
+
+
+class GBTDAALModel(GBTDAALBaseModel):
+    def __init__(self):
+        pass
+
+    def predict(self, X):
+        fptype = getFPType(X)
+        if self._is_regression:
+            return self._predict_regression(X, fptype)
+        else:
+            return self._predict_classification(X, fptype, "computeClassLabels")
+
+    def predict_proba(self, X):
+        fptype = getFPType(X)
+        if self._is_regression:
+            raise NotImplementedError("Can't predict probabilities for regression task")
+        else:
+            return self._predict_classification(X, fptype, "computeClassProbabilities")
+
+
+def convert_model(model):
+    gbm = GBTDAALModel()
+    gbm._convert_model(model)
+
+    gbm._is_regression = isinstance(gbm.daal_model_, d4p.gbt_regression_model)
+
+    return gbm

daal4py/sklearn/ensemble/GBTDAAL.py

@@ -27,7 +27,7 @@
 from .._utils import getFPType
 
 
-class GBTDAALBase(BaseEstimator):
+class GBTDAALBase(BaseEstimator, d4p.mb.GBTDAALBaseModel):
     def __init__(self,
                  split_method='inexact',
                  max_iterations=50,
@@ -101,6 +101,11 @@ def _check_params(self):
             raise ValueError('Parameter "min_bin_size" must be '
                              'non-zero positive integer value.')
 
+    allow_nan_ = False
+
+    def _more_tags(self):
+        return {"allow_nan": self.allow_nan_}
+
 
 class GBTDAALClassifier(GBTDAALBase, ClassifierMixin):
     def fit(self, X, y):
@@ -165,41 +170,28 @@ def fit(self, X, y):
         return self
 
     def _predict(self, X, resultsToEvaluate):
+        # Input validation
+        if not self.allow_nan_:
+            X = check_array(X, dtype=[np.single, np.double])
+        else:
+            X = check_array(X, dtype=[np.single, np.double], force_all_finite='allow-nan')
+
         # Check is fit had been called
         check_is_fitted(self, ['n_features_in_', 'n_classes_'])
 
-        # Input validation
-        X = check_array(X, dtype=[np.single, np.double])
-        if X.shape[1] != self.n_features_in_:
-            raise ValueError('Shape of input is different from what was seen in `fit`')
-
         # Trivial case
         if self.n_classes_ == 1:
             return np.full(X.shape[0], self.classes_[0])
 
-        if not hasattr(self, 'daal_model_'):
-            raise ValueError((
-                "The class {} instance does not have 'daal_model_' attribute set. "
-                "Call 'fit' with appropriate arguments before using this method.").format(
-                    type(self).__name__))
-
-        # Define type of data
         fptype = getFPType(X)
-
-        # Prediction
-        predict_algo = d4p.gbt_classification_prediction(
-            fptype=fptype,
-            nClasses=self.n_classes_,
-            resultsToEvaluate=resultsToEvaluate)
-        predict_result = predict_algo.compute(X, self.daal_model_)
+        predict_result = self._predict_classification(X, fptype, resultsToEvaluate)
 
         if resultsToEvaluate == "computeClassLabels":
             # Decode labels
             le = preprocessing.LabelEncoder()
             le.classes_ = self.classes_
-            return le.inverse_transform(
-                predict_result.prediction.ravel().astype(np.int64, copy=False))
-        return predict_result.probabilities
+            return le.inverse_transform(predict_result)
+        return predict_result
 
     def predict(self, X):
         return self._predict(X, "computeClassLabels")
@@ -218,6 +210,14 @@ def predict_log_proba(self, X):
 
         return proba
 
+    def convert_model(model):
+        gbm = GBTDAALClassifier()
+        gbm._convert_model(model)
+
+        gbm.classes_ = model.classes_
+        gbm.allow_nan_ = True
+        return gbm
+
 
 class GBTDAALRegressor(GBTDAALBase, RegressorMixin):
     def fit(self, X, y):
@@ -264,25 +264,21 @@ def fit(self, X, y):
         return self
 
     def predict(self, X):
-        # Check is fit had been called
-        check_is_fitted(self, ['n_features_in_'])
-
         # Input validation
-        X = check_array(X, dtype=[np.single, np.double])
-        if X.shape[1] != self.n_features_in_:
-            raise ValueError('Shape of input is different from what was seen in `fit`')
+        if not self.allow_nan_:
+            X = check_array(X, dtype=[np.single, np.double])
+        else:
+            X = check_array(X, dtype=[np.single, np.double], force_all_finite='allow-nan')
 
-        if not hasattr(self, 'daal_model_'):
-            raise ValueError((
-                "The class {} instance does not have 'daal_model_' attribute set. "
-                "Call 'fit' with appropriate arguments before using this method.").format(
-                    type(self).__name__))
+        # Check is fit had been called
+        check_is_fitted(self, ['n_features_in_'])
 
-        # Define type of data
         fptype = getFPType(X)
+        return self._predict_regression(X, fptype)
 
-        # Prediction
-        predict_algo = d4p.gbt_regression_prediction(fptype=fptype)
-        predict_result = predict_algo.compute(X, self.daal_model_)
+    def convert_model(model):
+        gbm = GBTDAALRegressor()
+        gbm._convert_model(model)
 
-        return predict_result.prediction.ravel()
+        gbm.allow_nan_ = True
+        return gbm

daal4py/sklearn/ensemble/__init__.py

@@ -19,5 +19,5 @@
 from .GBTDAAL import (GBTDAALClassifier, GBTDAALRegressor)
 from .AdaBoostClassifier import AdaBoostClassifier
 
-__all__ = ['RandomForestClassifier', 'RandomForestRegressor', 'GBTDAALClassifier',
-           'GBTDAALRegressor', 'AdaBoostClassifier']
+__all__ = ['RandomForestClassifier', 'RandomForestRegressor',
+           'GBTDAALClassifier', 'GBTDAALRegressor', 'AdaBoostClassifier']

doc/daal4py/examples.rst

@@ -25,9 +25,9 @@ General usage
 
 Building models from Gradient Boosting frameworks
 
-- `XGBoost* model conversion <https://github.com/intel/scikit-learn-intelex/blob/master/examples/daal4py/gbt_cls_model_create_from_xgboost_batch.py>`_
-- `LightGBM* model conversion <https://github.com/intel/scikit-learn-intelex/blob/master/examples/daal4py/gbt_cls_model_create_from_lightgbm_batch.py>`_
-- `CatBoost* model conversion <https://github.com/intel/scikit-learn-intelex/blob/master/examples/daal4py/gbt_cls_model_create_from_catboost_batch.py>`_
+- `XGBoost* model conversion <https://github.com/intel/scikit-learn-intelex/blob/master/examples/daal4py/model_builders_xgboost.py>`_
+- `LightGBM* model conversion <https://github.com/intel/scikit-learn-intelex/blob/master/examples/daal4py/model_builders_lightgbm.py>`_
+- `CatBoost* model conversion <https://github.com/intel/scikit-learn-intelex/blob/master/examples/daal4py/model_builders_catboost.py>`_
 
 
 Principal Component Analysis (PCA) Transform