[英]Convert a vector to a mask matrix using numpy
Assume we have the following vector: 假设我们有以下向量:
v = np.array([4, 0, 1])
The goal is to create the 5 x 3 matrix M
as follows: 目的是创建5 x 3矩阵
M
,如下所示:
[[0 1 0]
[0 0 1]
[0 0 0]
[0 0 0]
[1 0 0]]
Only one element in each column is equal to 1 for the corresponding index in v
. v
相应索引的每一列中只有一个元素等于1。 For example, since v[0]
is 4 then M[4, 0] == 1
, and since v[2]
is 1 then M[1, 2] == 1
. 例如,由于
v[0]
为4,则M[4, 0] == 1
,由于v[2]
为1,则M[1, 2] == 1
。
How can I build such a matrix in Python using scipy and numpy? 如何使用scipy和numpy在Python中建立这样的矩阵? In MATLAB you can do this with the
sparse
and full
functions in a single line. 在MATLAB中,您可以在一行中使用
sparse
函数和full
函数来执行此操作。 I'd prefer not to use a for
loop since I am looking for a vectorized implementation of this. 我不希望使用
for
循环,因为我正在寻找对此的矢量化实现。
You can do: 你可以做:
from scipy import sparse
inds = np.array([4, 0, 1])
values = np.ones_like(inds) # [1, 1, 1]
index = np.arange(inds.shape[0]) # 3
m = sparse.csc_matrix((values, (inds, index)), shape=(5, 3))
Output: 输出:
>>> m.todense()
matrix([[0, 1, 0],
[0, 0, 1],
[0, 0, 0],
[0, 0, 0],
[1, 0, 0]])
If you want a dense array output, you could just use two integer arrays to index the rows/cols of the nonzero elements: 如果需要密集的数组输出,则可以使用两个整数数组来索引非零元素的行/列:
v = np.array([4, 0, 1])
x = np.zeros((5, 3), np.int)
x[v, np.arange(3)] = 1
print(x)
# [[0 1 0]
# [0 0 1]
# [0 0 0]
# [0 0 0]
# [1 0 0]]
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.