方差阈值显示 RuntimeWarning: Degrees of freedom <= 0 for slice

Question

我有一个特征数据框，我想删除低方差特征。我使用以下函数过滤掉低方差列：

def variance_threshold_selector(data, threshold):

    selector = VarianceThreshold(threshold)
    selector.fit(data)

return data[data.columns[selector.get_support(indices=True)]]

对于不同的阈值，我收到以下警告：

/anaconda3/envs/Observation/lib/python3.7/site-packages/sklearn/feature_selection/_variance_threshold.py:77: RuntimeWarning: Degrees of freedom <= 0 for slice.
  self.variances_ = np.nanvar(X, axis=0)

/anaconda3/envs/Observation/lib/python3.7/site-packages/sklearn/feature_selection/_variance_threshold.py:88: RuntimeWarning: invalid value encountered in less_equal
  (self.variances_ <= self.threshold)):

/anaconda3/envs/Observation/lib/python3.7/site-packages/sklearn/feature_selection/_variance_threshold.py:99: RuntimeWarning: invalid value encountered in greater

这些警告是什么，我该如何解决它们？

Answer 1

关注此 post：

RuntimeWarning: Degrees of freedom <= 0 for slice

当你使用错误的形状时出现

关注此：

RuntimeWarning: invalid value encountered in less_equal

最有可能发生的原因是 np.nan 某处涉及的输入

然后是 post:

RuntimeWarning: invalid value encountered in greater

您的问题是由 out_vec 数组中的 NaN 或 Inf 元素引起的

方差阈值显示 RuntimeWarning: Degrees of freedom <= 0 for slice

Variance thresholding shows RuntimeWarning: Degrees of freedom <= 0 for slice

python

variance

scikit-learn