Question

我有一个numpy数组y [1,2,3,4,5,6,1,2,3,4,5,6]

然后我有一个numpy csr格式的矩阵X.

1）我需要在y中屏蔽元素6。然后，我需要屏蔽X中的相应行。

所以，y.shape是12.应该是10.X是12,20。应该是10,20。

如何在numpy

中执行此操作

Answer 1

如果您的'蒙面'输出y应该小于您的输入，那么您实际上并没有使用蒙版。

根据previous question的建议，您可以轻松找到y与6不同的指数

condition = (y != 6)

可以用来检索非{6}的y值的abool数组

y = y[condition]

您可以使用相同的condition来获取X的相应行，但它是CSR，因此不支持花哨的索引格式。你仍然可以将它转换为LIL然后再回来。

您还可以获取y !=6与

的索引

(indices,) = np.nonzero(y != 6)

这是一个常规整数数组，可用于索引X。

Answer 2

我不喜欢回答我自己的问题，但正确的解决办法是处理csr矩阵：

X = X[np.where(y != 6)[0]]
y = y[y != 6]

Answer 3

我不是100％确定我理解你的问题，但也许这会有所帮助：

>>> import numpy as np
>>> a = np.array(range(1,7)*2)  #Your array.
>>> a
array([1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6])
>>> b = np.array(list(a)*20).reshape(12,20) #just some matrix of right size and shape. np.empty(12,20) would probably work just as well.
>>> b
array([[1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2],
       [3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4],
       [5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6],
       [1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2],
       [3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4],
       [5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6],
       [1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2],
       [3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4],
       [5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6],
       [1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2],
       [3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4],
       [5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6]])


>>> mask = a != 6 #mask.  True for all points except ones where value == 6.
>>> b[mask,:]   #take points along first axis where mask==True, all points along second axis.
array([[1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2],
       [3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4],
       [5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6],
       [1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2],
       [3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4],
       [1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2],
       [3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4],
       [5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6],
       [1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2],
       [3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4, 5, 6, 1, 2, 3, 4]])

Answer 4

使用nonzero。

（假设已经定义了x）

import numpy as np
y = np.array([1,2,3,4,5,6,1,2,3,4,5,6])
keepers = np.nonzero(y != 6)
y = y[keepers]
x = x[keepers, :]

按值屏蔽数组然后屏蔽相应的矩阵

4 个答案: