AttributeError: 'MultiOutputClassifier' object has no attribute 'classes_'

Question

我想获得每个输出的每个 class 的预测概率。但是 classes_ 属性在 MultiOutputClassifier 上不存在。

如何将 classes 与输出相关联？

from sklearn.datasets import make_classification
from sklearn.multioutput import MultiOutputClassifier
from sklearn.ensemble import RandomForestClassifier
from sklearn.utils import shuffle
import numpy as np

X, y1 = make_classification(n_samples=16, n_features=8, n_informative=4, n_classes=4, random_state=1)
y2 = shuffle(y1, random_state=1)
Y = np.vstack((y1, y2)).T

forest = RandomForestClassifier(n_estimators=16, random_state=1)
multi_target_forest = MultiOutputClassifier(forest, n_jobs=-1)
multi_target_forest.fit(X, Y).predict(X)

multi_target_forest.predict_proba(X)
multi_target_forest.classes_

AttributeError: 'MultiOutputClassifier' object has no attribute 'classes_'

Answer 1

使用 MultiOutputClassifier 的 estimators_ 属性访问估算器属性。

试试这个！

['y_{} class_{}'.format(idx,klass) for idx,forest in enumerate(multi_target_forest.estimators_) \
                                    for klass in forest.classes_]


#output:
['y_0 class_0',
 'y_0 class_1',
 'y_0 class_2',
 'y_0 class_3',
 'y_1 class_0',
 'y_1 class_1',
 'y_1 class_2',
 'y_1 class_3']

AttributeError: 'MultiOutputClassifier' object has no attribute 'classes_'

AttributeError: 'MultiOutputClassifier' object has no attribute 'classes_'

python

classification

scikit-learn

multilabel-classification