[英]How to slice and extend a 2D numpy array?
I have a numpy array of size nxm
. 我有一个大小为
nxm
的numpy数组。 I want the number of columns to be limited to k
and rest of the columns to be extended in new rows. 我希望将列数限制为
k
,将其余列扩展为新行。 Following is the scenario - 以下是情景 -
Initial array: nxm
初始数组:
nxm
Final array: pxk
最终数组:
pxk
where p = (m/k)*n
其中
p = (m/k)*n
Eg. 例如。
n = 2, m = 6, k = 2
Initial array: 初始数组:
[[1, 2, 3, 4, 5, 6,],
[7, 8, 9, 10, 11, 12]]
Final array: 最终阵列:
[[1, 2],
[7, 8],
[3, 4],
[9, 10],
[5, 6],
[11, 12]]
I tried using reshape
but not getting the desired result. 我尝试使用
reshape
但没有得到所需的结果。
Here's one way to do it 这是一种方法
q=array([[1, 2, 3, 4, 5, 6,],
[7, 8, 9, 10, 11, 12]])
r=q.T.reshape(-1,2,2)
s=r.swapaxes(1,2)
t=s.reshape(-1,2)
as a one liner, 作为一个班轮,
q.T.reshape(-1,2,2).swapaxes(1,2).reshape(-1,2)
array([[ 1, 2],
[ 7, 8],
[ 3, 4],
[ 9, 10],
[ 5, 6],
[11, 12]])
EDIT: for the general case, use 编辑:对于一般情况,使用
q=arange(1,1+n*m).reshape(n,m) #example input
r=q.T.reshape(-1,k,n)
s=r.swapaxes(1,2)
t=s.reshape(-1,k)
one liner is: 一个班轮是:
q.T.reshape(-1,k,n).swapaxes(1,2).reshape(-1,k)
example for n=3,m=12,k=4
n=3,m=12,k=4
例子n=3,m=12,k=4
q=array([[ 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12],
[13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24],
[25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36]])
result is 结果是
array([[ 1, 2, 3, 4],
[13, 14, 15, 16],
[25, 26, 27, 28],
[ 5, 6, 7, 8],
[17, 18, 19, 20],
[29, 30, 31, 32],
[ 9, 10, 11, 12],
[21, 22, 23, 24],
[33, 34, 35, 36]])
Using numpy.vstack
and numpy.hsplit
: 使用
numpy.vstack
和numpy.hsplit
:
a = np.array([[1, 2, 3, 4, 5, 6,],
[7, 8, 9, 10, 11, 12]])
n, m, k = 2, 6, 2
np.vstack(np.hsplit(a, m/k))
result array: 结果数组:
array([[ 1, 2],
[ 7, 8],
[ 3, 4],
[ 9, 10],
[ 5, 6],
[11, 12]])
UPDATE As flebool commented , above code is very slow, because hsplit
returns a python list, and then vstack
reconstructs the final array from a list of arrays. 更新正如flebool评论的那样 ,上面的代码非常慢,因为
hsplit
返回一个python列表,然后vstack
从一个数组列表重建最终的数组。
Here's alternative solution that is much faster. 这是替代解决方案,速度更快。
a.reshape(-1, m/k, k).transpose(1, 0, 2).reshape(-1, k)
or 要么
a.reshape(-1, m/k, k).swapaxes(0, 1).reshape(-1, k)
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.