DecisionTree does not handle properly missing values in criterion partitioning

### Describe the bug

I tried using `RFECV` with `RandomForestClassifier` in version 1.4.0 on data containing NaNs and got the following error:
```
ValueError: Input contains NaN.
```
This is my first time opening an issue to an open-source project before, so I apologize if this is ill-formatted or lacking of details. Please let me know if I can provide more information.

### Steps/Code to Reproduce

```python
import numpy as np
import sklearn
from sklearn.datasets import load_iris
from sklearn.feature_selection import RFECV
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split

X, y = load_iris(as_frame=True, return_X_y=True)

rng = np.random.RandomState(42)
X_missing = X.copy()
mask = rng.binomial(n=np.array([1, 1, 1, 1]).reshape(1, -1),
                    p=(X['petal length (cm)'] / 8).values.reshape(-1, 1)).astype(bool)
X_missing[mask] = np.NaN

X_train, X_test, y_train, y_test = train_test_split(X_missing, y, random_state=13)

clf = RandomForestClassifier()
selector = RFECV(clf, cv=3)

selector.fit(X_train, y_train)
```

### Expected Results

I would expect no error since `RandomForestClassifier` supports NaNs and according to the documentation for `RFECV`,
![image](https://github.com/scikit-learn/scikit-learn/assets/87620495/baee7fe8-689b-4d48-aea4-ee56ea3e2b05)

For instance, the following code works just fine:
```python
clf.fit(X_train, y_train)
y_pred = clf.predict(X_test)
```

### Actual Results

```python
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[32], [line 14](vscode-notebook-cell:?execution_count=32&line=14)
     [11](vscode-notebook-cell:?execution_count=32&line=11) clf = RandomForestClassifier()
     [12](vscode-notebook-cell:?execution_count=32&line=12) selector = RFECV(clf, cv=3)
---> [14](vscode-notebook-cell:?execution_count=32&line=14) selector.fit(X_train, y_train)

File [c:\Users\eugen\anaconda3\envs\ufc\lib\site-packages\sklearn\base.py:1351](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/base.py:1351), in _fit_context.<locals>.decorator.<locals>.wrapper(estimator, *args, **kwargs)
   [1344](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/base.py:1344)     estimator._validate_params()
   [1346](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/base.py:1346) with config_context(
   [1347](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/base.py:1347)     skip_parameter_validation=(
   [1348](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/base.py:1348)         prefer_skip_nested_validation or global_skip_validation
   [1349](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/base.py:1349)     )
   [1350](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/base.py:1350) ):
-> [1351](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/base.py:1351)     return fit_method(estimator, *args, **kwargs)

File [c:\Users\eugen\anaconda3\envs\ufc\lib\site-packages\sklearn\feature_selection\_rfe.py:746](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:746), in RFECV.fit(self, X, y, groups)
    [743](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:743)     parallel = Parallel(n_jobs=self.n_jobs)
    [744](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:744)     func = delayed(_rfe_single_fit)
--> [746](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:746) scores = parallel(
    [747](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:747)     func(rfe, self.estimator, X, y, train, test, scorer)
    [748](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:748)     for train, test in cv.split(X, y, groups)
    [749](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:749) )
    [751](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:751) scores = np.array(scores)
    [752](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:752) scores_sum = np.sum(scores, axis=0)

File [c:\Users\eugen\anaconda3\envs\ufc\lib\site-packages\sklearn\feature_selection\_rfe.py:747](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:747), in <genexpr>(.0)
    [743](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:743)     parallel = Parallel(n_jobs=self.n_jobs)
    [744](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:744)     func = delayed(_rfe_single_fit)
    [746](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:746) scores = parallel(
--> [747](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:747)     func(rfe, self.estimator, X, y, train, test, scorer)
    [748](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:748)     for train, test in cv.split(X, y, groups)
    [749](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:749) )
    [751](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:751) scores = np.array(scores)
    [752](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:752) scores_sum = np.sum(scores, axis=0)

File [c:\Users\eugen\anaconda3\envs\ufc\lib\site-packages\sklearn\feature_selection\_rfe.py:35](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:35), in _rfe_single_fit(rfe, estimator, X, y, train, test, scorer)
     [33](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:33) X_train, y_train = _safe_split(estimator, X, y, train)
     [34](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:34) X_test, y_test = _safe_split(estimator, X, y, test, train)
---> [35](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:35) return rfe._fit(
     [36](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:36)     X_train,
     [37](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:37)     y_train,
     [38](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:38)     lambda estimator, features: _score(
     [39](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:39)         # TODO(SLEP6): pass score_params here
     [40](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:40)         estimator,
     [41](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:41)         X_test[:, features],
     [42](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:42)         y_test,
     [43](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:43)         scorer,
     [44](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:44)         score_params=None,
     [45](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:45)     ),
     [46](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:46) ).scores_

File [c:\Users\eugen\anaconda3\envs\ufc\lib\site-packages\sklearn\feature_selection\_rfe.py:308](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:308), in RFE._fit(self, X, y, step_score, **fit_params)
    [305](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:305) estimator.fit(X[:, features], y, **fit_params)
    [307](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:307) # Get importance and rank them
--> [308](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:308) importances = _get_feature_importances(
    [309](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:309)     estimator,
    [310](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:310)     self.importance_getter,
    [311](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:311)     transform_func="square",
    [312](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:312) )
    [313](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:313) ranks = np.argsort(importances)
    [315](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_rfe.py:315) # for sparse case ranks is matrix

File [c:\Users\eugen\anaconda3\envs\ufc\lib\site-packages\sklearn\feature_selection\_base.py:238](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_base.py:238), in _get_feature_importances(estimator, getter, transform_func, norm_order)
    [236](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_base.py:236) elif transform_func == "square":
    [237](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_base.py:237)     if importances.ndim == 1:
--> [238](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_base.py:238)         importances = safe_sqr(importances)
    [239](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_base.py:239)     else:
    [240](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/feature_selection/_base.py:240)         importances = safe_sqr(importances).sum(axis=0)

File [c:\Users\eugen\anaconda3\envs\ufc\lib\site-packages\sklearn\utils\__init__.py:773](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/__init__.py:773), in safe_sqr(X, copy)
    [757](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/__init__.py:757) def safe_sqr(X, *, copy=True):
    [758](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/__init__.py:758)     """Element wise squaring of array-likes and sparse matrices.
    [759](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/__init__.py:759) 
    [760](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/__init__.py:760)     Parameters
   (...)
    [771](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/__init__.py:771)          Return the element-wise square of the input.
    [772](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/__init__.py:772)     """
--> [773](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/__init__.py:773)     X = check_array(X, accept_sparse=["csr", "csc", "coo"], ensure_2d=False)
    [774](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/__init__.py:774)     if issparse(X):
    [775](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/__init__.py:775)         if copy:

File [c:\Users\eugen\anaconda3\envs\ufc\lib\site-packages\sklearn\utils\validation.py:1003](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:1003), in check_array(array, accept_sparse, accept_large_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, ensure_min_samples, ensure_min_features, estimator, input_name)
    [997](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:997)     raise ValueError(
    [998](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:998)         "Found array with dim %d. %s expected <= 2."
    [999](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:999)         % (array.ndim, estimator_name)
   [1000](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:1000)     )
   [1002](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:1002) if force_all_finite:
-> [1003](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:1003)     _assert_all_finite(
   [1004](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:1004)         array,
   [1005](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:1005)         input_name=input_name,
   [1006](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:1006)         estimator_name=estimator_name,
   [1007](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:1007)         allow_nan=force_all_finite == "allow-nan",
   [1008](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:1008)     )
   [1010](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:1010) if copy:
   [1011](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:1011)     if _is_numpy_namespace(xp):
   [1012](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:1012)         # only make a copy if `array` and `array_orig` may share memory`

File [c:\Users\eugen\anaconda3\envs\ufc\lib\site-packages\sklearn\utils\validation.py:126](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:126), in _assert_all_finite(X, allow_nan, msg_dtype, estimator_name, input_name)
    [123](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:123) if first_pass_isfinite:
    [124](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:124)     return
--> [126](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:126) _assert_all_finite_element_wise(
    [127](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:127)     X,
    [128](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:128)     xp=xp,
    [129](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:129)     allow_nan=allow_nan,
    [130](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:130)     msg_dtype=msg_dtype,
    [131](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:131)     estimator_name=estimator_name,
    [132](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:132)     input_name=input_name,
    [133](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:133) )

File [c:\Users\eugen\anaconda3\envs\ufc\lib\site-packages\sklearn\utils\validation.py:175](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:175), in _assert_all_finite_element_wise(X, xp, allow_nan, msg_dtype, estimator_name, input_name)
    [158](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:158) if estimator_name and input_name == "X" and has_nan_error:
    [159](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:159)     # Improve the error message on how to handle missing values in
    [160](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:160)     # scikit-learn.
    [161](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:161)     msg_err += (
    [162](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:162)         f"\n{estimator_name} does not accept missing values"
    [163](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:163)         " encoded as NaN natively. For supervised learning, you might want"
   (...)
    [173](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:173)         "#estimators-that-handle-nan-values"
    [174](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:174)     )
--> [175](file:///C:/Users/eugen/anaconda3/envs/ufc/lib/site-packages/sklearn/utils/validation.py:175) raise ValueError(msg_err)

ValueError: Input contains NaN.
```

### Versions

```shell
System:
    python: 3.9.18 (main, Sep 11 2023, 14:09:26) [MSC v.1916 64 bit (AMD64)]
executable: c:\Users\eugen\anaconda3\envs\ufc\python.exe
   machine: Windows-10-10.0.19045-SP0

Python dependencies:
      sklearn: 1.4.0
          pip: 23.0.1
   setuptools: 66.0.0
        numpy: 1.26.0
        scipy: 1.11.3
       Cython: 3.0.0
       pandas: 1.5.3
   matplotlib: 3.6.2
       joblib: 1.2.0
threadpoolctl: 2.2.0

Built with OpenMP: True

threadpoolctl info:
       filepath: C:\Users\eugen\anaconda3\envs\ufc\Library\bin\mkl_rt.2.dll
         prefix: mkl_rt
       user_api: blas
   internal_api: mkl
        version: 2022.1-Product
    num_threads: 8
threading_layer: intel

       filepath: C:\Users\eugen\anaconda3\envs\ufc\vcomp140.dll
         prefix: vcomp
       user_api: openmp
   internal_api: openmp
        version: None
    num_threads: 16
```


Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

DecisionTree does not handle properly missing values in criterion partitioning #28254

Describe the bug

Steps/Code to Reproduce

Expected Results

Actual Results

Versions

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Uh oh!

DecisionTree does not handle properly missing values in criterion partitioning #28254

Description

Describe the bug

Steps/Code to Reproduce

Expected Results

Actual Results

Versions

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions