Removing zero values from a numpy array of arrays

Question

Original datas

 [[ 0.00000000e+00  1.00000000e+00 -6.76207728e+00 -1.63236398e+01]
 [ 0.00000000e+00  1.00000000e+00  2.51283367e+01  1.13952157e+02]
 [ 0.00000000e+00  1.00000000e+00  3.11402956e+00 -5.16009612e+02]
 [ 0.00000000e+00  1.00000000e+00  3.10969787e+01  1.82175649e+02]
 [ 1.00000000e+00 -2.31269114e+00 -4.13720127e+02  3.55395844e+03]
 [ 1.00000000e+00  4.54598490e+01  6.19694322e+02  2.61091335e+03]
 [ 1.00000000e+00  7.36925014e-01 -4.49386738e+02 -1.22392549e+03]
 [ 1.00000000e+00  3.29511609e+00 -4.43413555e+02 -4.12677155e+03]]

I tried to remove zeros with this ode below

def removeZeroPadding(X):
  res = []
  for poly in enumerate(X):
      tmp = poly[1]
      tmp = tmp[tmp != 0]
      res.append(tmp)
  return res

This transforms into

[array([  1.        ,  -6.76207728, -16.32363975]), array([  1.        ,  25.1283367 , 113.95215706]), array([   1.        ,    3.11402956, -516.0096117 ]), array([  1.        ,  31.09697873, 182.17564943]), array([ 1.00000000e+00, -2.31269114e+00, -4.13720127e+02,  3.55395844e+03]), array([1.00000000e+00, 4.54598490e+01, 6.19694322e+02, 2.61091335e+03]), array([ 1.00000000e+00,  7.36925014e-01, -4.49386738e+02, -1.22392549e+03]), array([ 1.00000000e+00,  3.29511609e+00, -4.43413555e+02, -4.12677155e+03])]

How can I keep the structures of the original data with no zeros? Thanks

edit: it should look like this

 [[ 1.00000000e+00 -6.76207728e+00 -1.63236398e+01]
 [ 1.00000000e+00  2.51283367e+01  1.13952157e+02]
 [ 1.00000000e+00  3.11402956e+00 -5.16009612e+02]
 [ 1.00000000e+00  3.10969787e+01  1.82175649e+02]
 [ 1.00000000e+00 -2.31269114e+00 -4.13720127e+02  3.55395844e+03]
 [ 1.00000000e+00  4.54598490e+01  6.19694322e+02  2.61091335e+03]
 [ 1.00000000e+00  7.36925014e-01 -4.49386738e+02 -1.22392549e+03]
 [ 1.00000000e+00  3.29511609e+00 -4.43413555e+02 -4.12677155e+03]]

Do you want to look for columns consisting of all 0's and remove those? — shmulvad
– shmulvad, Commented Jun 14, 2020 at 14:11
With “keep the structures”, you mean you want to keep the initial dimensions of your array (i.e. 4 by 6 in your example)? If so, what values should take the place of your zeros? None? — hlzl
– hlzl, Commented Jun 14, 2020 at 14:12

hlzl · Accepted Answer · 2020-06-14 14:44:04Z

1

It seems like you do not want a list of arrays as the output but instead want a 2-dimensional array (i.e. [[...]] instead of [array([...]), array([...])]).

However, this is not possible as the rows of your array end up with different sizes after you trim them by removing the zeros (i.e. some end up having 3 elements and some have 4). If you want it to be a single array, same as for matrices in mathematics, all columns (and all rows) will need to have the same number of elements.

As an alternative, you could assign the elements you want to trim a different value, e.g. None, or you can create an empty array and fill it.

edited Jun 14, 2020 at 14:44

answered Jun 14, 2020 at 14:22

hlzl

5282 gold badges6 silver badges19 bronze badges

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

Comments

shmulvad · Accepted Answer · 2020-06-14 14:39:44Z

1

What you seek is fundamentally impossible. You can't both remove a variable number of entries from each row and at the same time be guaranted that each row has the same number of entries (which is required to get a ndarray of type [[...]]. hlzl's answer explains this in more detail.

This following function will remove 0-entries from the left and right side of each row, but what it returns will be of type [array([...]), array([...])]:

def trim_zeros_in_rows(ndarr):
    return np.array([np.trim_zeros(row) for row in ndarr])

print(trim_zeros_in_rows(np.array([[0., 0., 1., 0.],
                                   [2., 3., 4., 0.]])))

Prints:

[array([1.]) array([2., 3., 4.])]

answered Jun 14, 2020 at 14:39

shmulvad

6484 silver badges17 bronze badges

Comments

Laurent R · Accepted Answer · 2020-06-14 14:15:16Z

0

Have a look at https://numpy.org/doc/stable/reference/generated/numpy.trim_zeros.html trim_zeros.

It will avoid the for loop

answered Jun 14, 2020 at 14:15

Laurent R

8421 gold badge9 silver badges25 bronze badges

Collectives™ on Stack Overflow

Removing zero values from a numpy array of arrays

3 Answers 3

Comments

Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related