Numpy delete column out of 3d Array

Question

I have an 3d Numpy Array e.g.:

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

I want to remove the elements in the same column, e.g. column 1. So the result should be:

=[[[1]
   [3]]
  [[0]
   [6]]]

I tried the np.delete function:

print np.delete(array[:][:],1,axis=1)

But it dosen't produce the desired result, instead I get:

=[[[1 2]]
  [[0 5]]]

I found this in the np doc (Link):

np.delete(array, np.s_[::2], 1)

But this seems to only remove the uneven Numbers and I don't know how to modify it. Also the doc on the np.s Function did not take me any further.

Any Help would be highly appreciated!

Since this is a 3d-array, you should delete axis=2.

willeM_ Van Onsem
– willeM_ Van Onsem

2018-01-27 15:54:48 +00:00
Commented Jan 27, 2018 at 15:54 — willeM_ Van Onsem
– willeM_ Van Onsem, Commented Jan 27, 2018 at 15:54

willeM_ Van Onsem · Accepted Answer · 2018-01-27 16:21:10Z

2

Your array is a three dimensional array, so in case you want to delete the items with index 1 of the last index, you should use:

np.delete(array[:][:], 1, axis=2)

Note that you do not need the [:][:] part: this will have no effect, so you can use:

np.delete(array, 1, axis=2)

This will then generate:

>>> np.delete(array, 1, axis=2)
array([[[1],
        [3]],

       [[0],
        [6]]])

edited Jan 27, 2018 at 16:21

answered Jan 27, 2018 at 15:56

willeM_ Van Onsem

482k33 gold badges483 silver badges624 bronze badges

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

3 Comments

mLstudent33 Over a year ago

How would I delete just one element from each 2D array?

willeM_ Van Onsem Over a year ago

@mLstudent33: A numpy array can only store "rectangular" data, so how would you remove one element from a 2d array?

mLstudent33 Over a year ago

Van Onsen, thanks, I made a copy to copy into by deleting the column index for a row in a for loop instead.

Mohit Motwani · Accepted Answer · 2018-01-27 15:57:31Z

2

Deleting using axis= [0, 1] works while your working with 2d arrays. As the dimension increases, accordingly the axis value should also be increased. This should work.

np.delete(array[:][:], 1, axis=2)

answered Jan 27, 2018 at 15:57

Mohit Motwani

4,8124 gold badges21 silver badges50 bronze badges

Comments

kmario23 · Accepted Answer · 2018-01-27 17:00:22Z

You can very well use basic slicing since np.delete() doesn't actually delete anything from the original array. Instead it returns a copy (of what you're requesting).

In [242]: arr
Out[242]: 
array([[[1, 2],
        [3, 4]],

       [[0, 5],
        [6, 7]]])

# select both slices, select all rows but only first column
In [243]: arr[:, :, :1]
Out[243]: 
array([[[1],
        [3]],

       [[0],
        [6]]])

# same case as above but you can omit the `:`s with a single `...`
In [244]: arr[..., :1]
Out[244]: 
array([[[1],
        [3]],

       [[0],
        [6]]])

# sanity check for no copy made
In [245]: (arr[..., :1]).flags
Out[245]: 
  C_CONTIGUOUS : False
  F_CONTIGUOUS : False
  OWNDATA : False   # <=========== NO copy
  WRITEABLE : True
  ALIGNED : True
  UPDATEIFCOPY : False

However, as already mentioned, np.delete() returns the result as a copy.

In [246]: np.delete(arr[:][:], 1, axis=2).flags
Out[246]: 
  C_CONTIGUOUS : True
  F_CONTIGUOUS : False
  OWNDATA : True    # <========= Its own copy
  WRITEABLE : True
  ALIGNED : True
  UPDATEIFCOPY : False

However, if you're slicing from original array, you've to be careful while you make any changes to the result so that you don't accidentally change the original array.

Collectives™ on Stack Overflow

Numpy delete column out of 3d Array

3 Answers 3

3 Comments

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

3 Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related