Manipulation of numpy structured objects

Question

I've read a .mat file with scipy.io:

data = scipy.io.loadmat(etc.)

The function returns a dictionary, and my Matlab structure array is stored in a Numpy structured array. So far, so good. One of my fields is called repet1_F3, and should contain a vector of floats. I've accessed the vector using:

repet1_F3= data['repet1_F3']

repet1_F3 has a weird structure that I can't manipulate:

>>> repet1_F3
array(array([ 0.48856978,  0.74278461, 2.73300925,  2.72642893, 2.73684854,  2.74516561,  2.69143553]), dtype=object)

Am I doing something wrong? How could I convert this object into a standard numpy array?

What does this data look like in matlab?

Eric
– Eric

2016-04-13 01:28:43 +00:00
Commented Apr 13, 2016 at 1:28 — Eric
– Eric, Commented Apr 13, 2016 at 1:28
And what is data.shape?

Eric
– Eric

2016-04-13 01:29:25 +00:00
Commented Apr 13, 2016 at 1:29 — Eric
– Eric, Commented Apr 13, 2016 at 1:29

hpaulj · Accepted Answer · 2016-04-13 02:03:27Z

1

loadmat tends to wrap MATLAB structures in numpy object arrays.

array(array([ 0.48856978,  0.74278461, 
 2.73300925,  2.72642893,  2.73684854,  2.74516561,  2.69143553]), dtype=object)

Looks like a 1 element array of dtype object; that element is itself a 1d float array. The outer array probably has shape () (0d).

Try repet1_F3.item() or repet1_F3([]). One of those should give the inner array.

Do you know the MATLAB structure that contains numbers like this?

correction - it should be repet1_F3[()].

edited Apr 13, 2016 at 2:03

answered Apr 13, 2016 at 0:19

hpaulj

233k14 gold badges260 silver badges392 bronze badges

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

4 Comments

user1363251 Over a year ago

repet1_F3.item() worked, thank you. The Matlab structure is a structure array. Numbers displayed in Matlab have a 4 decimals precision, but they have more in python. Don't know why.

hpaulj Over a year ago

The 4 decimals is just a display setting; the actual values are 32 or 64 bit floats.

Eric Over a year ago

I think repet1_F3([]) should be repet1_F3[()]

hpaulj Over a year ago

you are right - the indexing [] go outside, the empty tuple inside. That's what happens when I don't have a live shell to test things in.

Collectives™ on Stack Overflow

Manipulation of numpy structured objects

1 Answer 1

4 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

4 Comments

Your Answer

Sign up or log in

Post as a guest

Related