Creating new field names within a new structure when trying to reshape data

I have a dataset saved into a structure which each person has several trials of 360 data points. The data is structured as variable1.pptID.data_CW. Where each person is a new field of pptID, and each person's data is split into two groups of trials (data_CW and data_CCW) where there are several trials in each group. For example:

variable1.pt15.dataCCW = transpose([1:360; 1:360; 1:360; 1:360; 1:360]);
variable1.pt15.dataCW = transpose([1:2:720; 1:2:720; 1:2:720; 1:2:720; 1:2:720]);

variable1.pt16.dataCCW = transpose([1:360; 1:360; 1:360; 1:360; 1:360; 1:360; 1:360; 1:360]);
variable1.pt16.dataCW = transpose([1:2:720; 1:2:720; 1:2:720; 1:2:720; 1:2:720; 1:2:720; 1:2:720; 1:2:720]);

I am trying to reshape each trial into 72 sections of 5 data points. If I was to do this for each column separately, I could run this as:

reshape((variable1.pt15.dataCCW(:,1)), 5, []);

However, I am trying to get this so I don't have to run it for every trial as this would be pretty inefficient! I have tried to run it within a for loop as follows:

pptID = ('pt15');

var1CCW = variable1.(pptID).data_CCW;
var1CW = variable1.(pptID).data_CW;
[m,n]=size(var1CCW);
for c_loop=1:n %do separately for each time series (i.e. each participant or each trial)
    splitCCW(:,c_loop)= reshape((var1CCW(:,c_loop)), 5, []);
    splitCW(:,c_loop)= reshape((var1CW(:,c_loop)), 5, []);
end

However, I found that as the reshape is a different size to the initial grouping of trials I realised I would have to make each trial into a separate field of a structure. I then tried the following:

pptID = ('pt15');

var1CCW = variable1.(pptID).data_CCW;
var1CW = variable1.(pptID).data_CW;
[m,n]=size(var1CCW);
for c_loop=1:n %do separately for each time series (i.e. each participant or each trial)
    splitCCW.(pptID).a({c_loop})= reshape((var1CCW(:,c_loop)), 5, []);
    splitCW.(pptID).a({c_loop})= reshape((var1CW(:,c_loop)), 5, []);
end

However, this still does not seem to work. I think my naming of the new fields of the structure is incorrect - but I can't figure out the right way of doing this! It keeps coming up with the following error message:

Unable to use a value of type cell as an index.

Any thoughts about why this might be would be appreciated, as well as ways to further improve my code. I may just be going about it in the complete wrong way! Please let me know if anything requires further clarification

Answer 1

[Matt J]

Your post is hard to follow, so I'm doing a bit of guessing as to what you want. Regardless, I think the data organization strategy needs some reconsidering, both before and after the reshaping.

I think first of all that your initial data should be organized as a struct array pt, where pt(i) is one person:

pt(1).CW=rand(360,4);
pt(1).CCW=rand(360,4);
...
pt(16).CW=rand(360,7);
pt(16).CCW=rand(360,7);

Also, I don't think you should be splitting up the trial data into separate fields or cells. Each matrix of trial data can be reshaped to a 3D array that is 5x72xN quite easily:.

for i=1:numel(pt)
    
    pt(i)=structfun(@(z)reshape(z,5,72,[])   , pt(i), 'uni',0);
    
end

[pkll201]

That's really useful, thank you. I haven't come across this way of structuring the data - and I think that it would work really well for this. Sorry it wasn't clear, I tried to put as much detail about the way I tried to solve the problem as I could! I think this approach should work well. I will go and have a go at restructuring the data :)

[pkll201]

Thanks for your suggestion, I definitely like the way of structuring the data! I just tried the above, and it comes up with the following error message:

Error using reshape

Product of known dimensions, 72, not divisible into total number of elements, 1.

Error in @(z)reshape(5,72,[])

I have tried to change a couple of things, but none of the changes have helped. Do you have any ideas where I might be going wrong here?

[Matt J]

My approach assumes all your fields are 360xN, whic was the case in your post.

[Steven Lord]

Why are you trying to reshape the scalar value 5 into a 72-by-something array? That wouldn't work unless something was 1/72 and you can't have an array with 1/72 columns in MATLAB.

Ah, it looks like Matt J missed the first input to the reshape call -- it should be z.

z = 1:(72*10);
size(z)

ans = 1×2

1 720

y = reshape(z, 5, 72, []); % Make y a 5-by-72-by-something
size(y) % To make y keep the same number of elements as z, something = 2

ans = 1×3

5 72 2

[Matt J]

Yep. I fixed it.

[pkll201]

Oh that's great, thank you so much!! That works perfectly. Thank you for all your help - you've come from a perspective I would have never thought of on my own :)

[Matt J]

@pkll201 You're quite welcome, but please Accept-click the answer to indicate that you are happy with it.