Why can't I get the shape of this numpy array?

Question

I am importing a matlab file using scipy.io, and trying to find its dimensions. It seems, even though the file is getting loaded into python, it's not able to give the dimensions. Why is that? And how to fix this?

>>> import scipy.io
>>> pref_mat = scipy.io.loadmat('pref_mat_loc.mat')
>>> R=pref_mat
>>> import numpy
>>> R=numpy.array(R)
>>> len(R)
0
>>> R # We see that the first line of the file is getting printed, means the file has been loaded.

array({'pref_mat': array([[0, 0, 0, ..., 0, 0, 0],
       [0, 0, 0, ..., 0, 0, 0],
       [0, 0, 0, ..., 0, 0, 0],
       ...,
       [0, 0, 0, ..., 0, 0, 0],
       [0, 0, 0, ..., 0, 0, 0],
       [0, 0, 0, ..., 0, 0, 0]], dtype=uint16), '__header__': b'MATLAB 5.0 MAT-file, Platform: PCWIN64, Created on: Mon May 08 23:42:05 2017', '__version__': '1.0', '__globals__': []}, dtype=object)

>>> len(R.shape) # But it appears here as though R is empty
0
>>> R.shape # As does here
()

len(R.shape) is the same as R.ndim for future reference. — Mad Physicist
– Mad Physicist, Commented May 9, 2017 at 16:28
What is the purpose of R=pref_mat? Why not just R = scipy.io.loadmat(...)? — Mad Physicist
– Mad Physicist, Commented May 9, 2017 at 16:29
Why in the name of ... are you doing R=numpy.array(R)? Do you understand what loadmat returns? — Mad Physicist
– Mad Physicist, Commented May 9, 2017 at 16:29
@MadPhysicist Why? What happened? I am not a regular python user, I just happpen to require it for this particular job. As I understand, loadmat loads the matlab file; I don't know what it returns though. Is it a dictionary? — Kristada673
– Kristada673, Commented May 9, 2017 at 16:35

hpaulj · Accepted Answer · 2017-05-09 22:47:43Z

3

pref_mat, returned by loadmat is a dictionary. You have wrapped it in an array, R. I can deduce the contents of pref_mat as the {} part of R:

{'pref_mat': array([[0, 0, 0, ..., 0, 0, 0],
       [0, 0, 0, ..., 0, 0, 0],
       [0, 0, 0, ..., 0, 0, 0],
       ...,
       [0, 0, 0, ..., 0, 0, 0],
       [0, 0, 0, ..., 0, 0, 0],
       [0, 0, 0, ..., 0, 0, 0]], dtype=uint16), 
'__header__': b'MATLAB 5.0 MAT-file, Platform: PCWIN64, Created on: Mon May 08 23:42:05 2017', 
'__version__': '1.0', 
'__globals__': []}

So the array you are interested in is

R = pref_mat['pref_mat']

That R should have the shape and dtype that you want. Though in summary view I only see 0's.

If the MATLAB had saved cells or structs the nesting of 'object' type arrays would get more complicated.

edited May 9, 2017 at 22:47

answered May 9, 2017 at 16:47

hpaulj

233k14 gold badges260 silver badges392 bronze badges

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

1 Comment

Kristada673 Over a year ago

Ok, thanks. It works. The data is ok, its a sparse matrix, hence the 0s. :)

Collectives™ on Stack Overflow

Why can't I get the shape of this numpy array?

1 Answer 1

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related