另一种使用矩阵的方法

>>> from numpy import matrix
>>> a = [ [1,2,3],[4,5,6],[7,8,9] ]
>>> matrix(a).transpose()[1].getA()[0]
array([2, 5, 8])
>>> matrix(a).transpose()[0].getA()[0]
array([1, 4, 7])

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]

尽管使用zip(*iterable)来转置一个嵌套列表，如果嵌套列表的长度不同，你也可以使用以下方法:

map(None, *[(1,2,3,), (4,5,), (6,)])

结果:

[(1, 4, 6), (2, 5, None), (3, None, None)]

第一列如下:

map(None, *[(1,2,3,), (4,5,), (6,)])[0]
#>(1, 4, 6)

你也可以用这个:

values = np.array([[1,2,3],[4,5,6]])
values[...,0] # first column
#[1,4]

注意:这对于内置数组和未对齐的数组无效(例如np.array([[1,2,3]，[4,5,6,7]]))

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]

[matrix[i][column] for i in range(len(matrix))]