模型评估：交叉验证法

 

K折交叉验证(K-fold cross-validation): 将样本分成K份，每份数量大致相等，然后用其他的某一份作为测试，其他样本作为训练集，得到一个模型和一组预测值及模型评估值；循环这个过程K次，得到K组模型评估值，对其取平均值即得到最终的评估结果

 

from sklearn.model_selection import cross_val_score

clf = svm.SVC(kernel='linear', C=1)

scores = cross_val_score(clf, iris.data, iris.target, cv=5)

scores  

>>> 

#5次交叉验证的得分                                         

array([ 0.96...,  1.  ...,  0.96...,  0.96...,  1.        ])

 

#这种数据切分方式可以打乱顺序

from sklearn.model_selection import cross_val_score

from sklearn.model_selection import ShuffleSplit

 

cv = ShuffleSplit(n_splits=3, test_size=0.3, random_state=0)

cross_val_score(clf, iris.data, iris.target, cv=cv)

 

通过交叉验证获取预测

from sklearn import metrics

rom sklearn.model_selection import cross_val_predict

 

predicted = cross_val_predict(clf, iris.data, iris.target, cv=10)

metrics.accuracy_score(iris.target, predicted)

>>> 

0.966

转载于:https://www.cnblogs.com/yongfuxue/p/10095414.html