numpy.delete not removing column from array

Question

I'm attempting to remove each column one at a time from an array and, based on the documentation and this question, thought the following should work:

print(all_input_data.shape)

for n in range(9):
    print(n)
    testArray = all_input_data.copy()
    print(testArray.shape)
    np.delete(testArray,[n],axis=1)
    print(testArray.shape)
    print(testArray[0:1][:])

The original matrix is all_input_data.

This is not causing any columns to be deleted or generating any other change to the array. The initial output for the snippet above is:

(682120, 9)

0

(682120, 9)

(682120, 9)

[[  2.37000000e+02   1.60000000e+01   9.90000000e+01   1.04910000e+03
    9.29000000e-01   9.86000000e-01   8.43000000e-01   4.99290000e+01
    1.97000000e+00]]

The delete command is not changing the shape of the matrix at all.

According to the documentation numpy.delete returns a copy with the specified column(s) deleted. Also note that n will be invalid for n >= 5 as you delete successively columns from the array. You should rather apply a while loop and always delete the first one. — a_guest
– a_guest, Commented Jul 12, 2017 at 7:29

MB-F · Accepted Answer · 2017-07-12 07:37:21Z

2

np.delete returns a copy of the input array with elements removed.

Return a new array with sub-arrays along an axis deleted.

There is no in place deletion of array elements in numpy.

Because np.delete returns a copy and does not modify the input there is no need to manually make a copy of all_input_data:

import numpy as np
all_input_data = np.random.rand(100, 9)

for n in range(9):
    print(n)
    testArray = np.delete(all_input_data,[n],axis=1)
    print(testArray.shape)
    print(testArray[0:1][:])

answered Jul 12, 2017 at 7:37

MB-F

23.8k5 gold badges71 silver badges127 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

MB-F Over a year ago

@a_guest I think you are wrong mainly for one reason: all_input_data is never modified. The rest follows from that.

MB-F Over a year ago

@a_guets You are wrong again. The array is different in each iteration. Maybe you should try to run the code before commenting. As I understood the question, the OP wants to delete columns one at a time, not successively.

MB-F Over a year ago

The original array is copied in every loop iteration, which strongly suggests an intent not to modify the array. But it's pretty useless to speculate on one's intent. Sooner or later the OP may clarify this point.

MB-F Over a year ago

Deleted comments make me look like talking to myself :)

David R Over a year ago

Thanks. You are correct in interpreting my intent, both in terms of the language used and the code.

user3053452 · Accepted Answer · 2017-07-12 07:54:15Z

1

From linked question consider this:

In [2]: a = np.arange(12).reshape(3,4)

In [3]: np.delete(a, [1,3], axis=1)
Out[3]:
array([[ 0,  2],
       [ 4,  6],
       [ 8, 10]])

In [4]: a
Out[4]:
array([[ 0,  1,  2,  3],
       [ 4,  5,  6,  7],
       [ 8,  9, 10, 11]])

In other words, if you want to save changes you should save to a new variable, but considering the size of your matrix this wouldn't be practical. What you could do is use ~~slice notation~~ indexing. It is explained here.

edited Jul 12, 2017 at 7:54

answered Jul 12, 2017 at 7:42

user3053452

6741 gold badge14 silver badges39 bronze badges

1 Comment

MB-F Over a year ago

Using slice notation (actually indexing) to remove individual columns requires advanced indexing which also copies the array just like delete.

Collectives™ on Stack Overflow

numpy.delete not removing column from array

2 Answers 2

5 Comments

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

5 Comments

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related