Running calculations over columns

I currently have a running sample entropy code that works over a 300 sample point window. This works fine for matrices that contain only one column. I was wondering how I could amend the script in a way that it will calculate the running sample entropy for each column of a matrix with multiple columns. I'm guessing maybe a for loop may work in some way?

I have included the code with a sample example of data

X = randi(100,10000,1);
OutPut = SampEn(2,0.2,X,300);

function SampEnN = SampEn(dim,r,data,windowSize)
%%Set-up
correl = zeros(1,2);
[N,M] = size(data);
SampEnN = nan(N,M);
for t = windowSize+1:N
 tolerance = std(data(t-windowSize:t, :),'omitnan')*r;
 dataT = data(t-windowSize:t,:);
 dataMat = zeros(dim+1,windowSize-dim);
 for i = 1:dim+1
   dataMat(i,:) = dataT(i:windowSize-dim+i-1);
 end
 for m = dim:dim+1
   count = zeros(1,windowSize-dim);
   tempMat = dataMat(1:m,:);
   for i = 1:windowSize-m
     % calculate Chebyshev distance, excluding self-matching case
     dist = max(abs(tempMat(:,i+1:windowSize-dim) - repmat(tempMat(:,i),1,windowSize-dim-i)));
     % calculate Heaviside function of the distance
     % User can change it to any other function
     % for modified sample entropy (mSampEn) calculation
     D = (dist < tolerance);
     count(i) = sum(D)/(windowSize-dim);
   end
   correl(m-dim+1) = sum(count)/(windowSize-dim);
 end
 C = log(correl(1)/correl(2));
 SampEnN(t,:) = C;
end
end

Answer 1

[Naman Chaturvedi]

Hi,

If you don't want to change the function, you can use the following for loop code to compute the desired output.

output=[];
for i=1:size(X,2)
    output=[output SampEn(2,0.2,X(:,i),300)];
end

Hope this helps.

You can also modify(generalize) your function to calculate for any 2d array. MATLAB inbuilt functions work along columns and hence can generate column-wise results for a matrix.

[Stephen23]

https://www.mathworks.com/help/matlab/matlab_prog/preallocating-arrays.html