1)AUC:metrics.roc_auc_score(y_true, y_score, average=’macro’, sample_weight=None, max_fpr=None))
y_true:测试集样本的真实类别标签;y_score:正样本的预测概率
2)ACC(准确率):accuracy_score(y_true, y_pred, normalize=True, sample_weight=None)
y_true:测试集样本的真实类别标签;y_pred:样本的预测结果标签。normalize如果为True,返回预测正确的概率值;如果为False,返回预测正确的样本数。
3)precision(精确率):precision_score(y_true, y_pred, labels=None, pos_label=1, average=’binary’, sample_weight=None)
4)recall(召回率):
recall_score(y_true, y_pred, labels=None, pos_label=1, average=’binary’, sample_weight=None)
5)f1_score:F1 = 2 * (precision * recall) / (precision + recall)
f1_score(y_true, y_pred, labels=None, pos_label=1, average=’binary’, sample_weight=None)
扩展:f_score:
precision_recall_fscore_support(y_true, y_pred, beta=1.0, labels=None, pos_label=1, average=None, warn_for=(‘precision’, ’recall’, ’f-score’), sample_weight=None)
F = (beta^2+1) * (precision * recall) / (precision + recall)
6)混淆矩阵:
confusion_matrix(y_true, y_pred, labels=None, sample_weight=None)
7)使用决策树或者随机森林分类后计算的特征重要性:
model.feature_importance_。特征重要性的取值顺序是模型训练样本的特征顺序。
模型评估方法metrics类、模型类常用属性
最新推荐文章于 2023-08-02 13:18:59 发布