ImportError: cannot import name 'parallel_helper'

Question

每当我运行 - from sklearn.ensemble import RandomForestClassifier

我遇到错误 - ImportError: cannot import name 'parallel_helper' 堆栈跟踪是 -

--------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-135-d80da5c856d8> in <module>()
      1 # feature removal using ROC-AUC score
----> 2 from sklearn.ensemble import RandomForestClassifier
      3 roc_values = []
      4 for feature in diabetes_MICE_X.columns:
      5     clf = RandomForestClassifier()

~/anaconda3/envs/python3/lib/python3.6/site-packages/sklearn/ensemble/__init__.py in <module>()
      5 
      6 from .base import BaseEnsemble
----> 7 from .forest import RandomForestClassifier
      8 from .forest import RandomForestRegressor
      9 from .forest import RandomTreesEmbedding

~/anaconda3/envs/python3/lib/python3.6/site-packages/sklearn/ensemble/forest.py in <module>()
     59 from ..exceptions import DataConversionWarning, NotFittedError
     60 from .base import BaseEnsemble, _partition_estimators
---> 61 from ..utils.fixes import parallel_helper, _joblib_parallel_args
     62 from ..utils.multiclass import check_classification_targets
     63 from ..utils.validation import check_is_fitted

ImportError: cannot import name 'parallel_helper'


Note - I'm using jupyter notebook (conda_python3) in sagemaker.
scipy version = 1.3.1
numpy version = 1.17.2
scikit version = 0.21.3 


one strange thing that i'm unable to figure out is - whenever i do 

import sklearn
sklearn.__version__

它给我的输出是 0.22

有人可以帮我解决这个问题吗？

Answer 1

最好的办法是重新启动内核。我在 google colab 上有过类似的经历，重启运行时解决了问题。

Answer 2

在 sagemaker 中克服这个问题的最好方法是使用生命周期配置。不要在 notebook 中执行 pip install，而是将所有 requirements.txt 安装写在生命周期配置中。笔记本将花费更多时间生成，但代码和库将是 pre-installed.

ImportError: cannot import name 'parallel_helper'

ImportError: cannot import name 'parallel_helper'

pip

python-3.x

scikit-learn

anaconda

amazon-sagemaker