python - 在 scikit learn 的交叉验证中使用混淆矩阵作为评分指标

Question

我正在 scikit learn 中创建一个管道，

pipeline = Pipeline([
    ('bow', CountVectorizer()),  
    ('classifier', BernoulliNB()), 
])

并使用交叉验证计算准确度

scores = cross_val_score(pipeline,  # steps to convert raw messages      into models
                     train_set,  # training data
                     label_train,  # training labels
                     cv=5,  # split data randomly into 10 parts: 9 for training, 1 for scoring
                     scoring='accuracy',  # which scoring metric?
                     n_jobs=-1,  # -1 = use all cores = faster
                     )

如何报告混淆矩阵而不是“准确度”？

Answer 1

您可以使用 cross_val_predict( See the scikit-learn docs ) 而不是 cross_val_score。

而不是做:

from sklearn.model_selection import cross_val_score
scores = cross_val_score(clf, x, y, cv=10)

你可以这样做:

from sklearn.model_selection import cross_val_predict
from sklearn.metrics import confusion_matrix
y_pred = cross_val_predict(clf, x, y, cv=10)
conf_mat = confusion_matrix(y, y_pred)