Python 如何删除numpy数组中的列？

How to remove a column in a numpy array?

提问by minerals

Imagine we have a 5x4 matrix. We need to remove only the first dimension. How can we do it with numpy?

想象一下，我们有一个 5x4 的矩阵。我们只需要删除第一个维度。我们如何用numpy做到这一点？

array([[  0.,   1.,   2.,   3.],
       [  4.,   5.,   6.,   7.],
       [  8.,   9.,  10.,  11.],
       [ 12.,  13.,  14.,  15.],
       [ 16.,  17.,  18.,  19.]], dtype=float32)

I tried:

我试过：

arr = np.arange(20, dtype=np.float32)
matrix = arr.reshape(5, 4)
new_arr = numpy.delete(matrix, matrix[:,0])
trimmed_matrix = new_arr.reshape(5, 3)

It looks a bit clunky. Am I doing it correctly? If yes, is there a cleaner way to remove the dimension without reshaping?

它看起来有点笨重。我做得对吗？如果是，是否有更清洁的方法来去除尺寸而不重塑？

采纳答案by Back2Basics

If you want to remove a column from a 2D Numpy array you can specify the columns like this

如果要从 2D Numpy 数组中删除一列，可以像这样指定列

to keep all rows and to get rid of column 0 (or start at column 1 through the end)

保留所有行并删除第 0 列（或从第 1 列开始直到结束）

a[:,1:]

another way you can specify the columns you want to keep ( and change the order if you wish) This keeps all rows and only uses columns 0,2,3

您可以指定要保留的列的另一种方法（并根据需要更改顺序）这会保留所有行并且仅使用列 0,2,3

a[:,[0,2,3]]

The documentation on this can be found here

可以在此处找到有关此的文档

And if you want something which specifically removes columns you can do something like this:

如果您想要专门删除列的内容，您可以执行以下操作：

idxs = list.range(4)
idxs.pop(2) #this removes elements from the list
a[:, idxs]

and @hpaulj brought up numpy.delete()

@hpaulj 提出了 numpy.delete()

This would be how to return a view of 'a' with 2 columns removed (0 and 2) along axis=1.

这将是如何返回沿轴 = 1 删除 2 列（0 和 2）的“a”视图。

np.delete(a,[0,2],1)

This doesn't actually remove the items from 'a', it's return value is a new numpy array.

这实际上并没有从 'a' 中删除项目，它的返回值是一个新的 numpy 数组。

回答by yevgeniy

You don't need the second reshape.

你不需要第二次重塑。

matrix=np.delete(matrix,0,1)

回答by hpaulj

The correct way to use deleteis to specify index and dimension, eg. remove the 1st (0) column (dimension 1):

正确的使用方法delete是指定索引和维度，例如。删除第一 (0) 列（维度 1）：

In [215]: np.delete(np.arange(20).reshape(5,4),0,1)
Out[215]: 
array([[ 1,  2,  3],
       [ 5,  6,  7],
       [ 9, 10, 11],
       [13, 14, 15],
       [17, 18, 19]])

other expressions that work:

其他有效的表达方式：

np.arange(20).reshape(5,4)[:,1:]
np.arange(20).reshape(5,4)[:,[1,2,3]]
np.arange(20).reshape(5,4)[:,np.array([False,True,True,True])]