我也对这个问题感兴趣，比较了速度

numpy.c_[a, a]
numpy.stack([a, a]).T
numpy.vstack([a, a]).T
numpy.ascontiguousarray(numpy.stack([a, a]).T)               
numpy.ascontiguousarray(numpy.vstack([a, a]).T)
numpy.column_stack([a, a])
numpy.concatenate([a[:,None], a[:,None]], axis=1)
numpy.concatenate([a[None], a[None]], axis=0).T

它们对任何输入向量a都做同样的事情。增长a的时间:

注意，所有不连续的变量(特别是stack/vstack)最终都比所有连续的变量快。如果需要连续，Column_stack(因为它的清晰性和速度)似乎是一个不错的选择。

代码重现情节:

import numpy as np
import perfplot

b = perfplot.bench(
    setup=np.random.rand,
    kernels=[
        lambda a: np.c_[a, a],
        lambda a: np.ascontiguousarray(np.stack([a, a]).T),
        lambda a: np.ascontiguousarray(np.vstack([a, a]).T),
        lambda a: np.column_stack([a, a]),
        lambda a: np.concatenate([a[:, None], a[:, None]], axis=1),
        lambda a: np.ascontiguousarray(np.concatenate([a[None], a[None]], axis=0).T),
        lambda a: np.stack([a, a]).T,
        lambda a: np.vstack([a, a]).T,
        lambda a: np.concatenate([a[None], a[None]], axis=0).T,
    ],
    labels=[
        "c_",
        "ascont(stack)",
        "ascont(vstack)",
        "column_stack",
        "concat",
        "ascont(concat)",
        "stack (non-cont)",
        "vstack (non-cont)",
        "concat (non-cont)",
    ],
    n_range=[2 ** k for k in range(23)],
    xlabel="len(a)",
)
b.save("out.png")

我觉得下面这些最优雅:

b = np.insert(a, 3, values=0, axis=1) # Insert values before column 3

插入的一个优点是它还允许您在数组中的其他位置插入列(或行)。此外，您可以轻松地插入整个向量，而不是插入单个值，例如复制最后一列:

b = np.insert(a, insert_index, values=a[:,2], axis=1)

这就导致:

array([[1, 2, 3, 3],
       [2, 3, 4, 4]])

对于时间，insert可能比JoshAdel的解决方案慢:

In [1]: N = 10

In [2]: a = np.random.rand(N,N)

In [3]: %timeit b = np.hstack((a, np.zeros((a.shape[0], 1))))
100000 loops, best of 3: 7.5 µs per loop

In [4]: %timeit b = np.zeros((a.shape[0], a.shape[1]+1)); b[:,:-1] = a
100000 loops, best of 3: 2.17 µs per loop

In [5]: %timeit b = np.insert(a, 3, values=0, axis=1)
100000 loops, best of 3: 10.2 µs per loop

有一个专门的函数。它被称为numpy.pad

a = np.array([[1,2,3], [2,3,4]])
b = np.pad(a, ((0, 0), (0, 1)), mode='constant', constant_values=0)
print b
>>> array([[1, 2, 3, 0],
           [2, 3, 4, 0]])

以下是它在文档字符串中所说的:

Pads an array.

Parameters
----------
array : array_like of rank N
    Input array
pad_width : {sequence, array_like, int}
    Number of values padded to the edges of each axis.
    ((before_1, after_1), ... (before_N, after_N)) unique pad widths
    for each axis.
    ((before, after),) yields same before and after pad for each axis.
    (pad,) or int is a shortcut for before = after = pad width for all
    axes.
mode : str or function
    One of the following string values or a user supplied function.

    'constant'
        Pads with a constant value.
    'edge'
        Pads with the edge values of array.
    'linear_ramp'
        Pads with the linear ramp between end_value and the
        array edge value.
    'maximum'
        Pads with the maximum value of all or part of the
        vector along each axis.
    'mean'
        Pads with the mean value of all or part of the
        vector along each axis.
    'median'
        Pads with the median value of all or part of the
        vector along each axis.
    'minimum'
        Pads with the minimum value of all or part of the
        vector along each axis.
    'reflect'
        Pads with the reflection of the vector mirrored on
        the first and last values of the vector along each
        axis.
    'symmetric'
        Pads with the reflection of the vector mirrored
        along the edge of the array.
    'wrap'
        Pads with the wrap of the vector along the axis.
        The first values are used to pad the end and the
        end values are used to pad the beginning.
    <function>
        Padding function, see Notes.
stat_length : sequence or int, optional
    Used in 'maximum', 'mean', 'median', and 'minimum'.  Number of
    values at edge of each axis used to calculate the statistic value.

    ((before_1, after_1), ... (before_N, after_N)) unique statistic
    lengths for each axis.

    ((before, after),) yields same before and after statistic lengths
    for each axis.

    (stat_length,) or int is a shortcut for before = after = statistic
    length for all axes.

    Default is ``None``, to use the entire axis.
constant_values : sequence or int, optional
    Used in 'constant'.  The values to set the padded values for each
    axis.

    ((before_1, after_1), ... (before_N, after_N)) unique pad constants
    for each axis.

    ((before, after),) yields same before and after constants for each
    axis.

    (constant,) or int is a shortcut for before = after = constant for
    all axes.

    Default is 0.
end_values : sequence or int, optional
    Used in 'linear_ramp'.  The values used for the ending value of the
    linear_ramp and that will form the edge of the padded array.

    ((before_1, after_1), ... (before_N, after_N)) unique end values
    for each axis.

    ((before, after),) yields same before and after end values for each
    axis.

    (constant,) or int is a shortcut for before = after = end value for
    all axes.

    Default is 0.
reflect_type : {'even', 'odd'}, optional
    Used in 'reflect', and 'symmetric'.  The 'even' style is the
    default with an unaltered reflection around the edge value.  For
    the 'odd' style, the extented part of the array is created by
    subtracting the reflected values from two times the edge value.

Returns
-------
pad : ndarray
    Padded array of rank equal to `array` with shape increased
    according to `pad_width`.

Notes
-----
.. versionadded:: 1.7.0

For an array with rank greater than 1, some of the padding of later
axes is calculated from padding of previous axes.  This is easiest to
think about with a rank 2 array where the corners of the padded array
are calculated by using padded values from the first axis.

The padding function, if used, should return a rank 1 array equal in
length to the vector argument with padded values replaced. It has the
following signature::

    padding_func(vector, iaxis_pad_width, iaxis, kwargs)

where

    vector : ndarray
        A rank 1 array already padded with zeros.  Padded values are
        vector[:pad_tuple[0]] and vector[-pad_tuple[1]:].
    iaxis_pad_width : tuple
        A 2-tuple of ints, iaxis_pad_width[0] represents the number of
        values padded at the beginning of vector where
        iaxis_pad_width[1] represents the number of values padded at
        the end of vector.
    iaxis : int
        The axis currently being calculated.
    kwargs : dict
        Any keyword arguments the function requires.

Examples
--------
>>> a = [1, 2, 3, 4, 5]
>>> np.pad(a, (2,3), 'constant', constant_values=(4, 6))
array([4, 4, 1, 2, 3, 4, 5, 6, 6, 6])

>>> np.pad(a, (2, 3), 'edge')
array([1, 1, 1, 2, 3, 4, 5, 5, 5, 5])

>>> np.pad(a, (2, 3), 'linear_ramp', end_values=(5, -4))
array([ 5,  3,  1,  2,  3,  4,  5,  2, -1, -4])

>>> np.pad(a, (2,), 'maximum')
array([5, 5, 1, 2, 3, 4, 5, 5, 5])

>>> np.pad(a, (2,), 'mean')
array([3, 3, 1, 2, 3, 4, 5, 3, 3])

>>> np.pad(a, (2,), 'median')
array([3, 3, 1, 2, 3, 4, 5, 3, 3])

>>> a = [[1, 2], [3, 4]]
>>> np.pad(a, ((3, 2), (2, 3)), 'minimum')
array([[1, 1, 1, 2, 1, 1, 1],
       [1, 1, 1, 2, 1, 1, 1],
       [1, 1, 1, 2, 1, 1, 1],
       [1, 1, 1, 2, 1, 1, 1],
       [3, 3, 3, 4, 3, 3, 3],
       [1, 1, 1, 2, 1, 1, 1],
       [1, 1, 1, 2, 1, 1, 1]])

>>> a = [1, 2, 3, 4, 5]
>>> np.pad(a, (2, 3), 'reflect')
array([3, 2, 1, 2, 3, 4, 5, 4, 3, 2])

>>> np.pad(a, (2, 3), 'reflect', reflect_type='odd')
array([-1,  0,  1,  2,  3,  4,  5,  6,  7,  8])

>>> np.pad(a, (2, 3), 'symmetric')
array([2, 1, 1, 2, 3, 4, 5, 5, 4, 3])

>>> np.pad(a, (2, 3), 'symmetric', reflect_type='odd')
array([0, 1, 1, 2, 3, 4, 5, 5, 6, 7])

>>> np.pad(a, (2, 3), 'wrap')
array([4, 5, 1, 2, 3, 4, 5, 1, 2, 3])

>>> def pad_with(vector, pad_width, iaxis, kwargs):
...     pad_value = kwargs.get('padder', 10)
...     vector[:pad_width[0]] = pad_value
...     vector[-pad_width[1]:] = pad_value
...     return vector
>>> a = np.arange(6)
>>> a = a.reshape((2, 3))
>>> np.pad(a, 2, pad_with)
array([[10, 10, 10, 10, 10, 10, 10],
       [10, 10, 10, 10, 10, 10, 10],
       [10, 10,  0,  1,  2, 10, 10],
       [10, 10,  3,  4,  5, 10, 10],
       [10, 10, 10, 10, 10, 10, 10],
       [10, 10, 10, 10, 10, 10, 10]])
>>> np.pad(a, 2, pad_with, padder=100)
array([[100, 100, 100, 100, 100, 100, 100],
       [100, 100, 100, 100, 100, 100, 100],
       [100, 100,   0,   1,   2, 100, 100],
       [100, 100,   3,   4,   5, 100, 100],
       [100, 100, 100, 100, 100, 100, 100],
       [100, 100, 100, 100, 100, 100, 100]])

np。插入也可以。

matA = np.array([[1,2,3], 
                 [2,3,4]])
idx = 3
new_col = np.array([0, 0])
np.insert(matA, idx, new_col, axis=1)

array([[1, 2, 3, 0],
       [2, 3, 4, 0]])

它沿着一个轴插入值，这里是new_col，在给定索引之前，这里是idx。换句话说，新插入的值将占据idx列，并将最初在idx处和之后的值向后移动。

我喜欢JoshAdel的回答，因为他关注的是表现。一个较小的性能改进是避免使用零进行初始化的开销，而这些初始化只会被覆盖。当N很大时，这有一个可测量的差异，用空代替零，零的列被写成一个单独的步骤:

In [1]: import numpy as np

In [2]: N = 10000

In [3]: a = np.ones((N,N))

In [4]: %timeit b = np.zeros((a.shape[0],a.shape[1]+1)); b[:,:-1] = a
1 loops, best of 3: 492 ms per loop

In [5]: %timeit b = np.empty((a.shape[0],a.shape[1]+1)); b[:,:-1] = a; b[:,-1] = np.zeros((a.shape[0],))
1 loops, best of 3: 407 ms per loop

在我的例子中，我必须向NumPy数组中添加一列1

X = array([ 6.1101, 5.5277, ... ])
X.shape => (97,)
X = np.concatenate((np.ones((m,1), dtype=np.int), X.reshape(m,1)), axis=1)

后 X.shape => (97,2)

array([[ 1. , 6.1101],
       [ 1. , 5.5277],
...