Plot data from nested structure

Hi.

I have a nested structure with multiple participants (n=17), multiple datasets, and multiple phases segmented out of each trial, ie:

metrics.sub(i).EDA(1,1) is the data for all participants for the mean of a time series from stage 1 for each participant.

There are six stages in each trial and I want to create a single figure with 6 boxplots comparing the mean of all participants across the 6 stages. So, this would be the data for each of the six boxplots:

metrics.sub(i).EDA(1,1);

metrics.sub(i).EDA(1,2);

metrics.sub(i).EDA(1,3);

metrics.sub(i).EDA(1,4);

metrics.sub(i).EDA(1,5);

metrics.sub(i).EDA(1,6);

I'm having trouble efficiently creating an array (or any other format the boxplot(x) will handle). Right now, my poor coding skills has me using four for loops for each metric and creating a new field in the structure for each phase of each trial. I need to create many of these plots.

Is there any way I can do this with a single for loop?

Sincerely,

Bungling in Connecticut

[Mathieu NOE]

maybe you should share some code and data

[Susan]

Honestly, my code is such a mess, you all might run away in terror!

The structure with all data is probably too big to share, but I've attached a smaller version. I'm hoping I may have a cleaner solution than what I've been doing in something like this, but it's hanging up right now (probably due to missing values:

%Create a 3D matrix to plot features across conditions 
boxplotmatrix = [];

for nsub = 1 : 17  %number of participants
    for nactivity = 1 : 6  %stages of the trial for each participant
        for nindex = 1 : 4  %metrics from the decomposed time series
           
            %This next line is probably not going to work??? Some participants
            %missing all data from EDA/ECG/EEG/etc, some just a single
            %activity, but it's different for each participant & measure.
            %These plots aren't for publication, just to view what we've
            %got
            if ~isempty(metrics.sub(nsub).EDA(nactivity, nindex))  
            end
            boxplotmatrix(nindex,nsub,nactivity) = metrics.sub(nsub).EDA(nactivity, nindex);
        end
    end
end

Unable to resolve the name 'metrics.sub'.

%Create all boxplots with one for loop
for nindex = 1 :  Nindices
    boxplot(squeeze(boxplotmatrix(nindex,:,:))); % Note the command squeeze!
    title(NameofIndex{nindex}) 
end

Answer 1

[Voss]

load Proj_metrics

Something like this may have been what you intended:

Nparticipants = numel(metrics.sub);

%Create a 3D matrix to plot features across conditions
boxplotmatrix = NaN(4,Nparticipants,6);
for nsub = 1:Nparticipants
    [nactivity,nindex] = size(metrics.sub(nsub).EDA);
    boxplotmatrix(1:nindex,nsub,1:nactivity) = permute(metrics.sub(nsub).EDA,[2 3 1]);
end

%Create all boxplots with one for loop
figure
Nindices = size(boxplotmatrix,1);
NameofIndex = "Index "+(1:Nindices);
for nindex = 1:Nindices
    subplot(Nindices,1,nindex)
    boxplot(permute(boxplotmatrix(nindex,:,:),[2 3 1]));
    title(NameofIndex{nindex}) 
end

Something like this may be useful as well:

Nparticipants = numel(metrics.sub);
measures = fieldnames(metrics.sub);
Nmeasures = numel(measures);

siz = zeros(Nparticipants,2,Nmeasures);
for mm = 1:Nmeasures
    for ii = 1:Nparticipants
        siz(ii,:,mm) = size(metrics.sub(ii).(measures{mm}));
    end
end
max_siz = max(siz,[],1);

C = cell(1,Nmeasures);
for mm = 1:Nmeasures
    %Create a 3D matrix to plot features across conditions 
    boxplotmatrix = NaN(max_siz(1,2,mm),Nparticipants,max_siz(1,1,mm));
    for ii = 1:Nparticipants
        boxplotmatrix(1:siz(ii,2,mm), ii, 1:siz(ii,1,mm)) = permute(metrics.sub(ii).(measures{mm}), [2 3 1]);
    end
    C{mm} = boxplotmatrix;
end

%Create all boxplots
figure('Position',[10 10 500 1200]);
Nindices_max = max(max_siz(1,2,:));
for mm = 1:Nmeasures
    Nindices = max_siz(1,2,mm);
    for nindex = 1:Nindices
        subplot(Nindices_max,Nmeasures,(nindex-1)*Nmeasures+mm);
        boxplot(permute(C{mm}(nindex,:,:),[2 3 1]));
        title(sprintf('%s: %d',measures{mm},nindex))
    end
end

[Susan]

Wow, wow, wow! Thank you so much!!! I was up into the wee hours fighting with this last night & you make it look so easy!

[Voss]

You're welcome!