adding values to numpy array to make them of equal shape

Question

I have a numpy array called 'MEL' of shape (94824,).

These values contain arrays of different shapes such as (99,13), (54, 13) (87, 13). I'd like to fill the arrays that are smaller than (99,13) with zeros or even better the mean value of that array.

MEL = numpy.ndarray and
for i in MEL: i = <class 'numpy.ndarray'> (i.shape = 99, 13) except for the ones that need to be filled
for j in i: j = <class 'numpy.ndarray'>

Up till now I have this:

max_len = np.max([len(a) for a in MEL])
for i in MEL:
    i = np.asarray([np.pad(a, (0, max_len - len(a)), 'constant', constant_values=0) for a in i])

But the shapes remain the same. any suggestions?

How can your numpy array MEL contain other arrays ? I guess Mel is a list of arrays. In that case, be carefull that len(a) only returns the length along the first dimension of the array. How do you want to fill the small arrays ? — Liris
– Liris, Commented Dec 14, 2018 at 13:48
I know it is a weird structure and i don't really know how to tackle it either I have added some information about it in the question — Eeuwigestudent1
– Eeuwigestudent1, Commented Dec 14, 2018 at 13:53
I actually do not understand how you created such a structure whitout encountering 'ValueError: setting an array element with a sequence.' — Liris
– Liris, Commented Dec 14, 2018 at 13:58
I didn't create the structure myself but imported it as an .npy file. — Eeuwigestudent1
– Eeuwigestudent1, Commented Dec 14, 2018 at 14:02

Liris · Accepted Answer · 2018-12-14 14:02:47Z

1

From what I understand of your question, MEL is a list of 94824 two dimensional arrays that have different shapes. You'd like to return arrays that have the same shape than the largest one, but filled with 0.

I guess the simplest it to create new arrays that have the appropriate shape and fill them with the former ones. A small example would be:

max_dim = [np.max([a.shape[0] for a in MEL]), np.max([a.shape[1] for a in MEL])]
new_MEL = []
for a in MEL:
    temp = np.zeros((max_dim[0], max_dim[1]))
    temp[:a.shape[0], :a.shape[1]] = a
    new_MEL.append(temp)

answered Dec 14, 2018 at 14:02

Liris

1,5463 gold badges16 silver badges35 bronze badges

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

1 Comment

Eeuwigestudent1 Over a year ago

Apparently you understood correctly cause it worked. Thank you!

Collectives™ on Stack Overflow

adding values to numpy array to make them of equal shape

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related