array = [[1,2,3,4],[5,6,7,8],[9,10,11,12],[13,14,15,16]]

col1 = [val[1] for val in array]
col2 = [val[2] for val in array]
col3 = [val[3] for val in array]
col4 = [val[4] for val in array]
print(col1)
print(col2)
print(col3)
print(col4)

Output:
[1, 5, 9, 13]
[2, 6, 10, 14]
[3, 7, 11, 15]
[4, 8, 12, 16]

如果你有一个数组

a = [[1, 2], [2, 3], [3, 4]]

然后像这样提取第一列:

[row[0] for row in a]

结果是这样的:

[1, 2, 3]

您是否使用了NumPy数组?Python有array模块，但不支持多维数组。普通的Python列表也是一维的。

然而，如果你有一个简单的二维列表，像这样:

A = [[1,2,3,4],
     [5,6,7,8]]

然后你可以像这样提取一个列:

def column(matrix, i):
    return [row[i] for row in matrix]

提取第二列(索引1):

>>> column(A, 1)
[2, 6]

或者简单地说:

>>> [row[1] for row in A]
[2, 6]

如果你在Python中有一个二维数组(不是numpy)，你可以像这样提取所有的列，

data = [
['a', 1, 2], 
['b', 3, 4], 
['c', 5, 6]
]

columns = list(zip(*data))

print("column[0] = {}".format(columns[0]))
print("column[1] = {}".format(columns[1]))
print("column[2] = {}".format(columns[2]))

执行这段代码会得到，

>>> print("column[0] = {}".format(columns[0]))
column[0] = ('a', 'b', 'c')

>>> print("column[1] = {}".format(columns[1]))
column[1] = (1, 3, 5)

>>> print("column[2] = {}".format(columns[2]))
column[2] = (2, 4, 6)

我更喜欢下一个提示: 将矩阵命名为matrix_a并使用column_number，例如:

import numpy as np
matrix_a = np.array([[1,2,3,4],[5,6,7,8],[9,10,11,12]])
column_number=2

# you can get the row from transposed matrix - it will be a column:
col=matrix_a.transpose()[column_number]

array = [[1,2,3,4],[5,6,7,8],[9,10,11,12],[13,14,15,16]]

col1 = [val[1] for val in array]
col2 = [val[2] for val in array]
col3 = [val[3] for val in array]
col4 = [val[4] for val in array]
print(col1)
print(col2)
print(col3)
print(col4)

Output:
[1, 5, 9, 13]
[2, 6, 10, 14]
[3, 7, 11, 15]
[4, 8, 12, 16]