python画ROC曲线并标记AUC

最新推荐文章于 2024-08-22 08:51:11 发布

mohana48833985

最新推荐文章于 2024-08-22 08:51:11 发布

阅读量1k

点赞数

本文链接：https://blog.csdn.net/Caiqiudan/article/details/108692561

版权

定义画图ROC函数

def roc(name): # 生成ROC曲线
    plt.figure(dpi=150,figsize=(3,3))
    plt.plot(fpr_test,tpr_test,'b:',label=f'测试集AUC={round(metrics.auc(fpr_test,tpr_test),4)}')
    plt.plot(fpr_train,tpr_train,'r-.',label=f'训练集AUC={round(metrics.auc(fpr_train,tpr_train),4)}')
    plt.legend();plt.title(f'ROC曲线-{name}');plt.xlabel('假正率');plt.ylabel('真正率')
    # plt.grid()
    plt.plot([0, 1], [0, 1], 'k--',linewidth = 0.5)

x2_train, x2_test, y2_train, y2_test = train_test_split(x1, y1, test_size=0.3,random_state=3)
from sklearn.linear_model import LogisticRegression as LR
model = LR()
model.fit(x2_train,y2_train)
# roc
fpr_test,tpr_test,th_test = metrics.roc_curve(y2_test,model.predict_proba(x2_test)[:,1]) # 测试集
fpr.append(fpr_test); tpr.append(tpr_test); auc.append(round(metrics.auc(fpr_test,tpr_test),4))
fpr_train,tpr_train,th_train = metrics.roc_curve(y2_train,model.predict_proba(x2_train)[:,1]) # 训练集
roc('逻辑回归')