ImportError：无法从“sklearn.utils”导入名称“safe_indexing”

Question

每当我尝试运行以下代码行时：

from imblearn.under_sampling import NearMiss

对于 Jupyter 笔记本上的欠采样（或过采样）不平衡数据，我收到此错误：

ImportError Traceback (most recent call last)
<ipython-input-21-c701341dce49> in <module>
----> 1 from imblearn.under_sampling import NearMiss

~\anaconda3\lib\site-packages\imblearn\under_sampling\__init__.py in <module>
        4 """
        5 
----> 6 from ._prototype_generation import ClusterCentroids
        7 
        8 from ._prototype_selection import RandomUnderSampler

~\anaconda3\lib\site-packages\imblearn\under_sampling\_prototype_generation\__init__.py in <module>
        4 """
        5 
----> 6 from ._cluster_centroids import ClusterCentroids
        7 
        8 __all__ = ['ClusterCentroids']

~\anaconda3\lib\site-packages\imblearn\under_sampling\_prototype_generation\_cluster_centroids.py in <module>
        15 from sklearn.cluster import KMeans
        16 from sklearn.neighbors import NearestNeighbors
---> 17 from sklearn.utils import safe_indexing
        18 
        19 from ..base import BaseUnderSampler

ImportError: cannot import name 'safe_indexing' from 'sklearn.utils' (C:\Users\anaconda3\lib\site-packages\sklearn\utils\__init__.py)

Answer 1

对于 imblearn.under_sampling，您是否尝试重新安装 package？：

pip install imbalanced-learn

康达：

conda install -c conda-forge imbalanced-learn

在 jupyter 笔记本中：

import sys
!{sys.executable} -m pip install <package_name>

If you have scikitlearn>=0.24 (as far as i see there is a dependency for imblearn now,as scikit-learn (>=0.23) https://imbalanced-learn.org/stable/install.html ) you may want to try:

try:
    from sklearn.utils import safe_indexing
except ImportError:
    from sklearn.utils import _safe_indexing

Answer 2

编辑..\Anaconda3\Lib\site-packages\sklearn\utils\ __init__.py 。

复制def _safe_indexing...直到下一个def并将代码重命名为def safe_indexing... 。

保持以前_safe_indexing 。

它以这种方式解决了这个问题。

Answer 3

在~\Anaconda3\Lib\site-packages\yellowbrick\classifier\threshold.py模块中替换：

from sklearn.utils import indexable, safe_indexing

和

from sklearn.utils import indexable, _safe_indexing

之后，重新启动 kernel。