sklearn.metrics.auc_auc值计算公式

from sklearn import cross_validation,metrics
from sklearn import svm

train_data,train_target = load(filename)#自定义加载数据函数，返回的是训练数据的数据项和标签项
train_x,test_x,train_y,test_y = cross_validation.train_test_split(train_data,train_target,test_size=0.2,random_state=27)#把训练集按0.2的比例划分为训练集和验证集
#start svm
clf = svm.SVC(C=5.0)
clf.fit(train_x,train_y)
predict_prob_y = clf.predict_proba(test_x)#基于SVM对验证集做出预测，prodict_prob_y 为预测的概率
#end svm ,start metrics 
test_auc = metrics.roc_auc_score(test_y,prodict_prob_y)#验证集上的auc值
print test_auc