python numpy 行 列个数_Python 用 numpy 随机抽样选择矩阵的多行或多列

numpy提供了多种随机选取的方法，这里放三种方法，效果相同。

以从一个矩阵中随机抽取n行为例，写了三个函数，粗鄙代码，见笑

import numpy as np

#=================================================================================

def rand_row1(array,dim_needed): #array为需要采样的矩阵，dim_needed为想要抽取的行数

row_total = array.shape[0]

row_sequence = np.arange(row_total)

np.random.shuffle(row_sequence)

return array[row_sequence[0:dim_needed],:]

#=================================================================================

def rand_row2(array,dim_needed): #array为需要采样的矩阵，dim_needed为想要抽取的行数

row_total = array.shape[0]

row_sequence = np.random.permutation(row_total)

return array[row_sequence[0:dim_needed],:]

#=================================================================================

def rand_row3(array,dim_needed): #array为需要采样的矩阵，dim_needed为想要抽取的行数

row_total = array.shape[0]

row_sequence= np.random.choice(row_total,dim_needed,replace=False, p=None)

return array[row_sequence,:]

#=================================================================================

#=============================测试

test_data = np.array([0, 0])

for i in range(100):

test_data = np.vstack((test_data, [i+1, i+1]))

# print(test_data)

batch = rand_row1(test_data,20)

# batch = rand_row2(test_data,20)

# batch = rand_row3(test_data,20)

print(batch.shape,batch)

随机抽取n列与之类似，同志，共勉！