For loop with different length

Hello,

I have a 1x10 cell (Var), where each cell is a 50*20000*A 3D array. I am trying to run the following code:

for k = 1:N
    for j = 1:50
        for i = 1:10
            test{i} = mean(Var{i}(j,:,k));
            test2{i} = std(Var{i}(j,:,k));
        end
    end
end

However, as apparent the number N will be different as A changes for each cell. How can I compute this with different N?

Answer 1

[Roofus Milton]

See last section of code. I used a counter variable which is incremented.

rows = 50;
columns = 20000;
randRange = [10, 50];

% create the cell array
data = cell(1, 10);

% loop over cell array and create matricies with varying size of third
% dimension
for p = 1:length(data)
    data{p} = rand(rows, columns, randi(randRange));
end

% get the size of each matrix
dimensions = cell2mat(cellfun(@size, data, 'UniformOutput', false)');

% calculate how many rows will be needed
totalRows = sum(dimensions(:, 1) .* dimensions(:, 3));

% preallocate the output matrix
output = zeros(totalRows, columns);

% initialize the counter variable
counter = 1;

% loop over cell
for i = 1:length(data)
    
    % loop over rows
    for j = 1:dimensions(i, 1)
        
        % loop over third dimension
        for k = 1:dimensions(i, 3)
            
            % assign the data
            output(counter, :) = data{i}(j, :, k);
            
            % increment the counter
            counter = counter + 1;
            
        end
    end
end

[Uerm]

Hi,

Thanks for the answer! It is very clear and understandable. However, the output variable will be a X*Y double. My idea was that I should do some calculations such as

m{i} = mean(data{i}(j,:,k))
s{i} = std(data{i}(j,:,k))

etc. where each cell should be a X*Y double. How can I do that?

[Roofus Milton]

This is really a different question. m, s, X, and Y are not defined in your code or mine. I believe the below accoplishes what you are getting at.

Replace the las two sections of code above with these snippets.

% preallocate the output matrix
outputData = zeros(totalRows, columns);

% create a cell array of functions you want to calculate
f = {@mean, @std};

% preallocate the stats matrix
outputStats = zeros(totalRows, length(f));

% initialize the counter variable
counter = 1;

% loop over cell
for i = 1:length(data)
    
    % loop over rows
    for j = 1:dimensions(i, 1)
        
        % loop over third dimension
        for k = 1:dimensions(i, 3)
            
            % assign the data
            outputData(counter, :) = data{i}(j, :, k);
            
            % loop over the stats you want to calculate
            for g = 1:length(f)
                outputStats(counter, g) = f{g}(outputData(counter, :));
            end
            
            % increment the counter
            counter = counter + 1;
            
        end
    end
end

output = {outputStats, outputData};

[Uerm]

Sorry, my bad. I have edited the original question. Thanks a lot, I will try the code to see if it works.