convertDatasetToMatlabSubspaceFormat.m

function convertedData = convertDatasetToMatlabSubspaceFormat(data, labels)
% ensure that the labels are categorical
labels = categorical(labels);
% find the uniqu labels in the training data
uniqueLabels = unique(labels);
numClasses = numel(uniqueLabels);
[numFeatures, ~] = size(data);
% Convert the integers 0 to 9 to categorical for comparison
categories = arrayfun(@(x) categorical(x), 0:(numClasses-1), 'UniformOutput', false);

% Use cellfun instead of arrayfun because we're now dealing with cell arrays
samplesPerClass = cellfun(@(x) sum(labels == x), categories);
% Find the max number of samples in any class to determine the size of the 3D array
maxSamples = max(samplesPerClass);

% Initialize the 3D matrix with zeros
newdata = zeros(numFeatures, maxSamples, numClasses);

% Fill the 3D matrix
for class = 0:(numClasses-1)
    classSamples = data(:, labels == categorical(class));
    % You can choose to replicate the data if there are fewer samples than maxSamples
    % or just fill the matrix with the available samples leaving zeros otherwise
    numClassSamples = size(classSamples, 2);
    newdata(:, 1:numClassSamples, class + 1) = classSamples;
    % If you want to replicate samples for classes with fewer samples, add that logic here
end
convertedData = newdata;
end