sklearn中的交叉验证(Cross-Validation)
sklearn是利用python进行机器学习中一个非常全面和好用的第三方库,用过的都说好。今天主要记录一下sklearn中关于交叉验证的各种用法,主要是对sklearn官方文档 Cross-validation: evaluating estimator performance进行讲解,英文水平好的建议读官方文档,里面的知识点很详细。
1. cross_val_score
对数据集进行指定次数的交叉验证并为每次验证效果评测其中,score 默认是以 scoring=’f1_macro’进行评测的,余外针对分类或回归还有:这需要from sklearn import metrics ,通过在cross_val_score 指定参数来设定评测标准;
当cv 指定为int 类型时,默认使用KFold 或StratifiedKFold 进行数据集打乱,下面会对KFold 和StratifiedKFold 进行介绍。
In [15]: from sklearn.model_selection import cross_val_scoreIn [16]: clf = svm.SVC(kernel='linear', C=1)In [17]: scores = cross_val_score(clf, iris.data, iris.target, cv=5)In [18]: scoresOut[18]: array([ 0.96666667, 1. , 0.96666667, 0.96666667, 1. ])In [19]: scores.mean()Out[19]: 0.98000000000000009
除使用默认交叉验证方式外,可以对交叉验证方式进行指定,如验证次数,训练集测试集划分比例等
In [20]: from sklearn.model_selection import ShuffleSplitIn [21]: n_samples = iris.data.shape[0]In [22]: cv = ShuffleSplit(n_splits=3, test_size=.3, random_state=0)In [23]: cross_val_score(clf, iris.data, iris.target, cv=cv)Out[23]: array([ 0.97777778, 0.97777778, 1. ])
2. cross_val_predict
cross_val_predict 与cross_val_score 很相像,不过不同于返回的是评测效果,cross_val_predict 返回的是estimator 的分类结果(或回归值),这个对于后期模型的改善很重要,可以通过该预测输出对比实际目标值,准确定位到预测出错的地方,为我们参数优化及问题排查十分的重要。
In [28]: from sklearn.model_selection import cross_val_predictIn [29]: from sklearn import metricsIn [30]: predicted = cross_val_predict(clf, iris.data, iris.target, cv=10)In [31]: predictedOut[31]: array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2])In [32]: metrics.accuracy_score(iris.target, predicted)Out[32]: 0.96666666666666667
3. KFold
K折交叉验证,这是将数据集分成K份的官方给定方案,所谓K折就是将数据集通过K次分割,使得所有数据既在训练集出现过,又在测试集出现过,当然,每次分割中不会有重叠。相当于无放回抽样。
In [33]: from sklearn.model_selection import KFoldIn [34]: X = ['a','b','c','d']In [35]: kf = KFold(n_splits=2)In [36]: for train, test in kf.split(X): ...: print train, test ...: print np.array(X)[train], np.array(X)[test] ...: print '\n' ...: [2 3] [0 1]['c' 'd'] ['a' 'b'][0 1] [2 3]['a' 'b'] ['c' 'd']
4. LeaveOneOut
LeaveOneOut 其实就是KFold 的一个特例,因为使用次数比较多,因此独立的定义出来,完全可以通过KFold 实现。In [37]: from sklearn.model_selection import LeaveOneOutIn [38]: X = [1,2,3,4]In [39]: loo = LeaveOneOut()In [41]: for train, test in loo.split(X): ...: print train, test ...: [1 2 3] [0][0 2 3] [1][0 1 3] [2][0 1 2] [3]#使用KFold实现LeaveOneOtutIn [42]: kf = KFold(n_splits=len(X))In [43]: for train, test in kf.split(X): ...: print train, test ...: [1 2 3] [0][0 2 3] [1][0 1 3] [2][0 1 2] [3]
5. LeavePOut
这个也是KFold 的一个特例,用KFold 实现起来稍麻烦些,跟LeaveOneOut 也很像。In [44]: from sklearn.model_selection import LeavePOutIn [45]: X = np.ones(4)In [46]: lpo = LeavePOut(p=2)In [47]: for train, test in lpo.split(X): ...: print train, test ...: [2 3] [0 1][1 3] [0 2][1 2] [0 3][0 3] [1 2][0 2] [1 3][0 1] [2 3]
6. ShuffleSplit
ShuffleSplit 咋一看用法跟LeavePOut 很像,其实两者完全不一样,LeavePOut 是使得数据集经过数次分割后,所有的测试集出现的元素的集合即是完整的数据集,即无放回的抽样,而ShuffleSplit 则是有放回的抽样,只能说经过一个足够大的抽样次数后,保证测试集出现了完成的数据集的倍数。In [48]: from sklearn.model_selection import ShuffleSplitIn [49]: X = np.arange(5)In [50]: ss = ShuffleSplit(n_splits=3, test_size=.25, random_state=0)In [51]: for train_index, test_index in ss.split(X): ...: print train_index, test_index ...: [1 3 4] [2 0][1 4 3] [0 2][4 0 2] [1 3]
7. StratifiedKFold
对测试集合进行无放回抽样
In [52]: from sklearn.model_selection import StratifiedKFoldIn [53]: X = np.ones(10)In [54]: y = [0,0,0,0,1,1,1,1,1,1]In [55]: skf = StratifiedKFold(n_splits=3)In [56]: for train, test in skf.split(X,y): ...: print train, test ...: [2 3 6 7 8 9] [0 1 4 5][0 1 3 4 5 8 9] [2 6 7][0 1 2 4 5 6 7] [3 8 9]
原文:https://blog.csdn.net/xiaodongxiexie/article/details/71915259