Deselecting a column without deletion in Numpy-CodePudding

I have a numpy array in this form:

n = [[[5 0 2]
  [8 9 7]
  [2 2 2]
  [5 9 5]   <-- target value
  [4 1 5]]
 [[5 3 9]
  [4 2 7]
  [7 0 7]
  [4 9 6]   <-- target value
  [3 8 5]]]

I want to get all the values except the 3rd row from each individual array. ie. the results should be in this way:

 [[[5 0 2]
  [8 9 7]
  [2 2 2]
  [4 1 5]]

 [[5 3 9]
  [4 2 7]
  [7 0 7]
  [3 8 5]]]

I tried using indexing, but I cannot achieve it. I have a large numpy array where this operation needs to be done. Is there any efficient method to do it, other than making a copy of it and removing it using delete method. Also, I don't want to utilize much space by creating a copy, I just want to use indexing to ignore the particular column for a while.

CodePudding user response：

You can use numpy.delete(arr, index, axis) like below:

>>> n
array([[[5, 0, 2],
        [8, 9, 7],
        [2, 2, 2],
        [5, 9, 5],
        [4, 1, 5]],

       [[5, 3, 9],
        [4, 2, 7],
        [7, 0, 7],
        [4, 9, 6],
        [3, 8, 5]]])

>>> np.delete(n, 3, 1)
array([[[5, 0, 2],
        [8, 9, 7],
        [2, 2, 2],
        [4, 1, 5]],

       [[5, 3, 9],
        [4, 2, 7],
        [7, 0, 7],
        [3, 8, 5]]])

Update: without delete:

>>> msk = (0,1,2,4)
>>> n[:,msk,:]

CodePudding user response：

You can create a list of selected elements, popping the target one:

import numpy as np

n = np.array([[[5,0,2],
               [8,9,7],
               [2,2,2],
               [5,9,5],   
               [4,1,5]],
              [[5,3,9],
               [4,2,7],
               [7,0,7],
               [4,9,6], 
               [3,8,5]]])

target_element = 3

s = list(range(len(n[0])))
s.pop(target_element)
print(n[:,s])

s = list(range(len(n[0])))
print(n[:,s[:target_element]   s[target_element 1:]])