K折交叉验证(KFold)

K折交叉验证：sklearn.model_selection.KFold(n_splits=n, shuffle=False, random_state=None)

思路：将训练/测试数据集划分为n个互斥子集，每次用其中一个子集当作验证集，剩下的n-1个作为训练集，进行n次训练和测试，得到n个结果

注：对于不能均等份的数据集，其前n_samples % n子集拥有int(n_samples / n)+ 1个样本，其余子集都只有int(n_samples / n)样本

参数说明：

n_splits：表示划分几等份

shuffle：在每次划分时，是否进行洗牌（默认为False）

      ①若为Falses时，其效果等同于random_state等于整数，每次划分的结果相同

      ②若为True时，每次划分的结果都不一样，表示经过洗牌，随机取样的

random_state：随机种子数

属性：

      ①get_n_splits(X=None, y=None, groups=None)：获取参数n_splits的值

      ②split(X, y=None, groups=None)：将数据集划分成训练集和测试集，返回索引生成器

实例：

      ①设置shuffle=False，运行两次结果相同

from sklearn.model_selection import KFold
import numpy as np
X = np.arange(20).reshape(5, 4)
y = np.random.randint(1, 2, 5)
# first
kf = KFold(n_splits=5)
for train_index, test_index in kf.split(X):
    print('train_index:%s , test_index: %s ' % (train_index, test_index))
    
train_index:[1 2 3 4] , test_index: [0] 
train_index:[0 2 3 4] , test_index: [1] 
train_index:[0 1 3 4] , test_index: [2] 
train_index:[0 1 2 4] , test_index: [3] 
train_index:[0 1 2 3] , test_index: [4] 


# second
kf = KFold(n_splits=5, shuffle=False)
for train_index, test_index in kf.split(X):
    print('train_index:%s , test_index: %s ' % (train_index, test_index))
    
train_index:[1 2 3 4] , test_index: [0] 
train_index:[0 2 3 4] , test_index: [1] 
train_index:[0 1 3 4] , test_index: [2] 
train_index:[0 1 2 4] , test_index: [3] 
train_index:[0 1 2 3] , test_index: [4] 

      ②设置shuffle=True时，两次运行的结果不同

from sklearn.model_selection import KFold
import numpy as np
X = np.arange(20).reshape(5, 4)
y = np.random.randint(1, 2, 5)
# first
kf = KFold(n_splits=5,shuffle=True)
for train_index, test_index in kf.split(X):
    print('train_index:%s , test_index: %s ' % (train_index, test_index))
    
train_index:[0 1 2 4] , test_index: [3] 
train_index:[0 2 3 4] , test_index: [1] 
train_index:[0 1 2 3] , test_index: [4] 
train_index:[0 1 3 4] , test_index: [2] 
train_index:[1 2 3 4] , test_index: [0] 


#second
kf = KFold(n_splits=5,shuffle=True)
for train_index, test_index in kf.split(X):
    print('train_index:%s , test_index: %s ' % (train_index, test_index))
    
train_index:[0 1 3 4] , test_index: [2] 
train_index:[0 2 3 4] , test_index: [1] 
train_index:[0 1 2 4] , test_index: [3] 
train_index:[0 1 2 3] , test_index: [4] 
train_index:[1 2 3 4] , test_index: [0] 

      ③设置shuffle=True和random_state=整数，每次运行的结果都相同

from sklearn.model_selection import KFold
import numpy as np
X = np.arange(20).reshape(5, 4)
y = np.random.randint(1, 2, 5)
# first
kf = KFold(n_splits=5,shuffle=True,random_state=1)
for train_index, test_index in kf.split(X):
    print('train_index:%s , test_index: %s ' % (train_index, test_index))
    
train_index:[0 1 3 4] , test_index: [2] 
train_index:[0 2 3 4] , test_index: [1] 
train_index:[0 1 2 3] , test_index: [4] 
train_index:[1 2 3 4] , test_index: [0] 
train_index:[0 1 2 4] , test_index: [3] 


# second
kf = KFold(n_splits=5,shuffle=True,random_state=1)
for train_index, test_index in kf.split(X):
    print('train_index:%s , test_index: %s ' % (train_index, test_index))
    
train_index:[0 1 3 4] , test_index: [2] 
train_index:[0 2 3 4] , test_index: [1] 
train_index:[0 1 2 3] , test_index: [4] 
train_index:[1 2 3 4] , test_index: [0] 
train_index:[0 1 2 4] , test_index: [3] 

​​​​​​​      ④n_splits属性值获取

kf.split(X)
<generator object _BaseKFold.split at 0x000002153CACE970>
kf.get_n_splits()
5
kf.n_splits
5

最后

以上就是烂漫鞋子为你收集整理的K折交叉验证(KFold)的全部内容，希望文章能够帮你解决K折交叉验证(KFold)所遇到的程序开发问题。

如果觉得靠谱客网站的内容还不错，欢迎将靠谱客网站推荐给程序员好友。