Average and plot variable from multiple .mat files

Hi! I am trying to average several variables contained in multiple .mat files. The variables I would like to perform a grand average on are 'ersp' and 'itc'. I would then also like to plot a figure for each of the average across all the files, with respect to the 'hemisphere' and the 'ROI'. However, when I try to take a squeeze(mean(ersp)) this does not work. The variables 'ersp' and 'itc' are arrays with values like (25 x 200 x 64) however, this last value changes depending on the files (it corresponds to the number of channels which is different per file).

Here is my script so far:

saving =[OPT.SAVEFOLDER];
mkdir(saving); cd(saving);
save(['DATA_' patient '_' hemisphere '_' timepoint '_' OPT.ROIS{roi} '.mat'],'EEG','ersp','itc','powbase','times','frequencies','-v7.3');                
myFolder = [OPT.SAVEFOLDER]; 
if ~isdir(myFolder)
   errorMessage = sprintf('Error: The following folder does not exist:\n%s', myFolder);
   uiwait(warndlg(errorMessage));
   return;
end
filePattern = fullfile(myFolder, 'DATA_*.mat');
matFiles = dir(filePattern);
for k = 1:length(matFiles)
    matFilename = fullfile(myFolder, matFiles(k).name)
    matData = load(matFilename);
end 
for k = 1:length(matFiles)
   matFilename = fullfile(myFolder, matFiles(k).name)
   matData = load(matFilename); 
   hasField = isfield(matData, 'ersp');
   if ~hasField
     warningMessage = sprintf('ersp is not in %s\n', matFilename);
     uiwait(warndlg(warningMessage));
     continue; 
   end
   ersp = matData.ersp; % Extract ERSP %                   
   figure; plot(squeeze(mean(ersp,'omitnan')));
end

Answer 1

[dpb]

The first loop over k below is superfluous...it just loads each file into the same array but does nothing with it; then you start all over again immediately. Just delete that loop entirely.

for k = 1:length(matFiles)
    matFilename = fullfile(myFolder, matFiles(k).name)
    matData = load(matFilename);
end 
for k = 1:length(matFiles)
   matFilename = fullfile(myFolder, matFiles(k).name)
   matData = load(matFilename); 
   ...

Then, you need two things -- an accumulator for the means across the number of files by variable and to compute the mean for the entire array if, as it sounds, that is the wanted statistic--that needs the (:) Matlab idiom to return the full array regardless of dimensions/size as a single (column) vector:

vars={'ersp','itc'};                                      % variables wanted from .mat file
nF=length(matFiles);                                      % how many files found?
mns=nan(nF,length(vars));                                 % allocate space; NaN so if missing won't plot 0
for k=1:nF
  matFilename = fullfile(myFolder, matFiles(k).name);     % fully-qualified file name
  s=load(matFilename vars{:});                            % load the desired variables in struct s
  mns(k,:)=structfun(@(x) mean(x(:),'omitnan')).';         % compute means by file
end

Leaves you with the array of means for the variables in vars requested to plot as desired.

[Kristin]

Thank you for your help!!

A few questions, so I can learn..

1. what is the mns function?
2. Does the last line work as is? When I ran it I got several errors, about the unexpected (i:) for example. I'm very new to this platform, so I'm not entirely sure how to fix it.

[dpb]

mns isn't a function; it's an array allocated to collect the computed means.

Oh, pooh! I see a couple of typos in the last line; the LHS needs to have a comma separating the row index from the column : and I pasted from command line to the editor and had used i instead of k as a variable. The line should be

mns(k,:)=structfun(@(x) mean(x(:),'omitnan')).';