fall back on unsupported bootstrapping beyond 100% (#2984)

Author: david-cortes-intel

doc/sources/algorithms.rst

@@ -55,6 +55,7 @@ Classification
        - ``ccp_alpha`` != `0`
        - ``criterion`` != `'gini'`
        - ``n_estimators`` > ``6024``
+       - Non-integer ``max_samples`` larger than 1 with ``bootstrap=True``
      - Multi-output and sparse data are not supported. Missing values and infinite values are not supported.
      - Number of classes must be at least 2.
    * - :obj:`sklearn.ensemble.ExtraTreesClassifier`
@@ -64,6 +65,7 @@ Classification
        - ``ccp_alpha`` != `0`
        - ``criterion`` != `'gini'`
        - ``n_estimators`` > ``6024``
+       - Non-integer ``max_samples`` larger than 1 with ``bootstrap=True``
      - Multi-output and sparse data are not supported. Missing values and infinite values are not supported.
      - Number of classes must be at least 2.
    * - :obj:`sklearn.neighbors.KNeighborsClassifier`
@@ -125,6 +127,7 @@ Regression
        - ``ccp_alpha`` != `0`
        - ``criterion`` != ``'squared_error'``
        - ``n_estimators`` > ``6024``
+       - Non-integer ``max_samples`` larger than 1 with ``bootstrap=True``
      - Multi-output and sparse data are not supported. Missing values and infinite values are not supported.
    * - :obj:`sklearn.ensemble.ExtraTreesRegressor`
      - All parameters are supported except:
@@ -133,6 +136,7 @@ Regression
        - ``ccp_alpha`` != `0`
        - ``criterion`` != ``'squared_error'``
        - ``n_estimators`` > ``6024``
+       - Non-integer ``max_samples`` larger than 1 with ``bootstrap=True``
      - Multi-output and sparse data are not supported. Missing values and infinite values are not supported.
    * - :obj:`sklearn.neighbors.KNeighborsRegressor`
      -
@@ -375,6 +379,7 @@ Classification
        - ``oob_score`` = `True`
        - ``sample_weight`` != `None`
        - ``n_estimators`` > ``6024``
+       - Non-integer ``max_samples`` larger than 1 with ``bootstrap=True``
      - Multi-output and sparse data are not supported. Missing values and infinite values are not supported.
      - Number of classes must be at least 2.
    * - :obj:`sklearn.ensemble.ExtraTreesClassifier`
@@ -386,6 +391,7 @@ Classification
        - ``oob_score`` = `True`
        - ``sample_weight`` != `None`
        - ``n_estimators`` > ``6024``
+       - Non-integer ``max_samples`` larger than 1 with ``bootstrap=True``
      - Multi-output and sparse data are not supported. Missing values and infinite values are not supported.
      - Number of classes must be at least 2.
    * - :obj:`sklearn.neighbors.KNeighborsClassifier`
@@ -429,6 +435,7 @@ Regression
        - ``oob_score`` = `True`
        - ``sample_weight`` != `None`
        - ``n_estimators`` > ``6024``
+       - Non-integer ``max_samples`` larger than 1 with ``bootstrap=True``
      - Multi-output and sparse data are not supported. Missing values and infinite values are not supported.
    * - :obj:`sklearn.ensemble.ExtraTreesRegressor`
      - All parameters are supported except:
@@ -439,6 +446,7 @@ Regression
        - ``oob_score`` = `True`
        - ``sample_weight`` != `None`
        - ``n_estimators`` > ``6024``
+       - Non-integer ``max_samples`` larger than 1 with ``bootstrap=True``
      - Multi-output and sparse data are not supported. Missing values and infinite values are not supported.
    * - :obj:`sklearn.neighbors.KNeighborsRegressor`
      - All parameters are supported except:
@@ -610,6 +618,7 @@ Classification
        - ``oob_score`` = `True`
        - ``sample_weight`` != `None`
        - ``n_estimators`` > ``6024``
+       - Non-integer ``max_samples`` larger than 1 with ``bootstrap=True``
      - Multi-output and sparse data are not supported. Missing values and infinite values are not supported.
      - Number of classes must be at least 2.
    * - :obj:`sklearn.ensemble.ExtraTreesClassifier`
@@ -621,6 +630,7 @@ Classification
        - ``oob_score`` = `True`
        - ``sample_weight`` != `None`
        - ``n_estimators`` > ``6024``
+       - Non-integer ``max_samples`` larger than 1 with ``bootstrap=True``
      - Multi-output and sparse data are not supported. Missing values and infinite values are not supported.
      - Number of classes must be at least 2.
    * - :obj:`sklearn.neighbors.KNeighborsClassifier`
@@ -665,6 +675,7 @@ Regression
        - ``oob_score`` = `True`
        - ``sample_weight`` != `None`
        - ``n_estimators`` > ``6024``
+       - Non-integer ``max_samples`` larger than 1 with ``bootstrap=True``
      - Multi-output and sparse data are not supported. Missing values and infinite values are not supported.
    * - :obj:`sklearn.ensemble.ExtraTreesRegressor`
      - All parameters are supported except:
@@ -675,6 +686,7 @@ Regression
        - ``oob_score`` = `True`
        - ``sample_weight`` != `None`
        - ``n_estimators`` > ``6024``
+       - Non-integer ``max_samples`` larger than 1 with ``bootstrap=True``
      - Multi-output and sparse data are not supported. Missing values and infinite values are not supported.
    * - :obj:`sklearn.neighbors.KNeighborsRegressor`
      - All parameters are supported except:

sklearnex/ensemble/_forest.py

@@ -19,6 +19,7 @@
 import warnings
 from abc import ABC
 from functools import partial
+from numbers import Integral, Real
 
 import numpy as np
 from sklearn.base import clone, is_classifier
@@ -285,6 +286,14 @@ def _onedal_fit_ready(self, patching_status, X, y, sample_weight):
                     not self.bootstrap or self.class_weight != "balanced_subsample",
                     "'balanced_subsample' for class_weight is not supported",
                 ),
+                (
+                    not (
+                        isinstance(self.max_samples, Real)
+                        and not isinstance(self.max_samples, Integral)
+                        and self.max_samples > 1.0
+                    ),
+                    "Fractional 'max_samples' beyond 1.0 are not supported",
+                ),
             ]
         )