如何对离散数据使用mutual_info_classif

Question

基本上，我试图使用下面的脚本来使用这个函数，以便了解一些有关离散特性的内容：

from sklearn.feature_selection import mutual_info_classif

mutual_info_classif(r1[to_consider].values, r1['Y'].values, discrete_features='True')

但是，它会引发一个错误：

---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
<ipython-input-52-c20422a21616> in <module>()
      1 from sklearn.feature_selection import mutual_info_classif
      2 
----> 3 mutual_info_classif(np.array(r1[to_consider].values), np.array(r1['Y'].values), discrete_features='True')

~\AppData\Local\Continuum\Anaconda3\lib\site-packages\sklearn\feature_selection\mutual_info_.py in mutual_info_classif(X, y, discrete_features, n_neighbors, copy, random_state)
    448     check_classification_targets(y)
    449     return _estimate_mi(X, y, discrete_features, True, n_neighbors,
--> 450                         copy, random_state)

~\AppData\Local\Continuum\Anaconda3\lib\site-packages\sklearn\feature_selection\mutual_info_.py in _estimate_mi(X, y, discrete_features, discrete_target, n_neighbors, copy, random_state)
    259         if discrete_features.dtype != 'bool':
    260             discrete_mask = np.zeros(n_features, dtype=bool)
--> 261             discrete_mask[discrete_features] = True
    262         else:
    263             discrete_mask = discrete_features

IndexError: arrays used as indices must be of integer (or boolean) type

  [1]: http://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.mutual_info_classif.html

我想弄明白为什么会发生这种事。以下是我的数据(热编码)的预览：

r1[to_consider].values
array([[ 0,  7,  1,  1,  2,  5],
       [ 1,  0,  1,  0,  0,  5],
       [ 0,  0,  1,  1,  6,  5],
       ...,
       [ 0,  0,  1,  1,  6,  3],
       [ 3, 11,  2,  2, 10,  5],
       [ 0,  0,  1,  1,  9,  0]], dtype=int64)

和：

r1['Y'].values
array([0, 0, 1, ..., 0, 0, 0], dtype=int8)

Answer 1

我假设你的

r1

是一个DataFrame。尝试这样做，而不是在值中提供信息。

from sklearn.feature_selection import mutual_info_classif

mutual_info_classif(r1[to_consider], r1['Y'], discrete_features='True')