How do I Average across structure

I have a 14 subject structure (ERP_data.ERP_data); for each subject there is 3D matrix: 32(electrode) x 381(data) x 32(condition); I would like to average the data across subjects with nanmean, so the final matrix will be 32x381x32; thank you for you help!

R

[madhan ravi]

Would you upload the data file/files and the code you tried?

[madhan ravi]

Where’s the data file?

[Rene Sebena]

Sorry, but datafile is too big to upload...you can see the structure of that file on the picture.

Answer 1

[Jacob Ward]

Here's a possible solution I used to solve your problem:

clear; close all;
% This part just creates dummy data similar to yours with the same structure
ERP_data(14) = struct('ERP_data',0);
for n = 1:14
    ERP_data(n) = struct('ERP_data',rand(32,381,32));
end

averagedData = zeros(32,381,32); % Initializes final matrix
for n = 1:32
    for m = 1:381
        for o = 1:32
            dataToAverage = zeros(14,1); % Initializes vector to average
            for p = 1:14
                dataToAverage(p) = ERP_data(p).ERP_data(n,m,o); % Grabs the data point from each subject for the position given by m,n, and o
            end
            averagedData(n,m,o) = nanmean(dataToAverage); % Averages those points and then assigns that value to the position given by m,n, and o in the final matrix
        end
    end
end

Hope that helps!

[Rene Sebena]

Thanks a lot! This is what I did meanwhile, and it is also working:

GA_ERP_cue = []; for ch = 1:32 GA_ERP_cue_temp = nanmean([ERP_data(1).ERP_data(ch,:,:); ERP_data(2).ERP_data(ch,:,:); ERP_data(3).ERP_data(ch,:,:); ERP_data(4).ERP_data(ch,:,:); ERP_data(5).ERP_data(ch,:,:); ERP_data(6).ERP_data(ch,:,:); ERP_data(7).ERP_data(ch,:,:); ERP_data(8).ERP_data(ch,:,:); ERP_data(9).ERP_data(ch,:,:); ERP_data(10).ERP_data(ch,:,:); ERP_data(11).ERP_data(ch,:,:); ERP_data(12).ERP_data(ch,:,:); ERP_data(13).ERP_data(ch,:,:); ERP_data(14).ERP_data(ch,:,:)]); GA_ERP_cue(ch,:,:) = GA_ERP_cue_temp; end

Answer 2

[Image Analyst]

You can use mean() with the 'omitnan' option.