average sections of matrix

Hi there, I am trying to average sections of a matrix at a time. The matrix is 31x109. Initially I am trying to build another matrix from the average of the first 25 elements increments in the first row, then the second 25(beginning from the the last element in the previous average); this will later be applied to all rows. My current attempt is

for f = 1:[1:25:108]; Ave_PpIX(f,:) = PpIX_Dose_Matrix(1,1:f); end

ave_PpIX = mean(Ave_PpIX);

Am idea of the matrix is: [ave(1-25) ave(13-38) ave(25-50)...]

If anyone has had any previous experience doing this I would really appreciate any help.

Thnaks Luke

[Jan]

Why 108 in the loop, not 109? What should happen with the last chunk, which has less then 25 elements?

1:[a:b:c] is the same as 1:a.

Answer 1

[Star Strider]

I am not certain that I understand the result you want.

Try this:

M = rand(31, 109);                                          % Original Matrix
Cols = [ones(1, fix(size(M,2)/25))*25 rem(size(M,2),25)];   % Create Column Sub-Sections 
C = mat2cell(M, size(M,1), Cols);                           % Cell Array Of Sub-Sections
Cmean = cellfun(@(x)mean(x,2), C, 'Uni',0);                 % Row Means For Each Section
Mmean = cell2mat(Cmean);                                    % Convert To Double Matrix Of Row Mean Vectors For Each Sub-Section

[Luke McLellan]

Thank you very much.

[Star Strider]

My pleasure.

‘I have just realised ... that the matrix values should be [1-25 13-38 25-50...]...’

You can easily implement this in my code by changing the ‘Cols’ vector, and segment the matrix as many times as necessary to get all the segments you want.

For example one line would be:

Cols1 = [25 25 ...];
C1 = mat2cell(M, size(M,1), Cols1);

and a second call:

Cols2 = [12 26 ...];                        % The Second Segment Will Be Columns 13-38
C2 = mat2cell(M, size(M,1), Cols2);

The code becomes a bit more complicated with the additional segmentation steps. It is still relatively straightforward and efficient. The column dimensions in the ‘Cols1’ and ‘Cols2’ vectors each have to total the column size (dimension 2) of your matrix, 109 here.

[Luke McLellan]

Thanks again. Still need to figure out the matrix segregation for Cols2.

[Star Strider]

As always, my pleasure.

If you only want columns 13-38, this works:

Cols2 = [12 26 size(M,2)-(12+26)];

The second segment of the mat2cell result is the one you want.

To be certain ‘Cols2’ (or any other segmentation vector) will work in mat2cell, ‘sum(Cols2)’ has to equal ‘size(M,2)’. (All elements must of course be integers.)

[Luke McLellan]

I had it so that I was repeating 3 iterations of 13-38 rather than following the trend but your code, I believe, follows the sequencing? Am I correct in saying that?

[Luke McLellan]

No, sorry. It doesn't continue the sequence.

[Star Strider]

The 13-38 segment is 26 elements, so to repeat 3 consecutive segments, ‘Cols2’ becomes:

Cols2 = [12 26 26 26 size(M,2)-(26*3+12)];

that becomes ‘[12 26 26 26 19]’, in mat2cell creating segments from columns 1-12. 13-38, 39-64, 65-90, and 91-109. So here, use segments 2, 3, and 4 out of the 5 segments created.

[Luke McLellan]

I do believe that you are my knight in shining algorithm but I will have to return to the code next week now

[Star Strider]

Thank you!

Answer 2

[Jan]

PpIX_Dose_Matrix = rand(31, 109);
n = size(PpIX_Dose_Matrix, 2);
k = 0;
for f = 1:24:n
  k = k + 1;
  Ave_PpIX(k) = mean(PpIX_Dose_Matrix(1, f:f+24));
end

This can fail, when f+25 is outside the existing range. A test with if can catch this:

if f+25 <= n
  ...

[Luke McLellan]

Thank you very much. I have implemented it. I have just realised (probably through my tiredness yesterday) that the matrix values should be [1-25 13-38 25-50...] so time to work on this haha.

[Jan]

for f = 1:12:n
  k = k + 1;
  Ave_PpIX(k) = mean(PpIX_Dose_Matrix(1, f:f+24));
end

[Luke McLellan]

Thanks again for your comments