Question

我可以使用warnings库通过scikit-learn通过几个选项关闭警告：

# After the imports
warnings.filterwarnings(action='ignore')
# Or in the code
with warnings.catch_warnings():
    warnings.simplefilter("ignore") 
    # do stuff

但是，一旦n_jobs参数大于1（由于要进行多处理？），则对分类器不起作用。以下代码示例对此进行了说明：

import numpy as np
from sklearn.multiclass import OneVsRestClassifier
from sklearn.linear_model import LogisticRegression
import warnings
import logging

logger = logging.getLogger()

for n_job in [1, 2]:
    print("START")
    print("n_jobs =", n_job)
    clf = OneVsRestClassifier(LogisticRegression(solver="liblinear", multi_class="ovr"), n_jobs=n_job)

    x_train = np.array([[1,1], [0,1], [0,0], [1,5], [2,1], [3,1]])
    y_train = np.array([[False, False, True], [False, False, True], [True, False, False], [True, False, False], [True, False, True], [True, False, False]])

    with warnings.catch_warnings():
        warnings.simplefilter("ignore")
        clf.fit(x_train, y_train) # "UserWarning: Label not 1 is present in all training examples."
    print("END")
    print()

输出：

START
n_jobs = 1
END

START
n_jobs = 2
UserWarning: Label not 1 is present in all training examples.
END

即使n_jobs> 1，如何禁用警告？

编辑：由于可能与multiprocessing相关，所以我可能要补充一点，我在Linux上使用python 3.6运行了该脚本。

使用n_jobs> 1时关闭scikit-learn的警告

0 个答案: