Creating a matrix where each row is the average of rows of another matrix.

Suppose I have 10*2 matrix A. I want to create 5*2 matrix B where the first row of A is the average of the first two rows of B, the second row of A is the average of the third and the fourth row of B, etc. Please advise.

Answer 1

[Andrei Bobrov]

One way:

a = [...
   102    20
   114   122
    16   120
   115    61
    80   101
    13    18
    35    53
    69   115
   120   100
   121   120];

out = splitapply(@mean,a,ceil((1:size(a,1))'/2));

other way:

[ii,jj] = ndgrid(ceil((1:size(a,1))'/2),1:size(a,2));
out = accumarray([ii(:),jj(:)],a(:),[],@mean);

third way:

n = mod(-size(a,1),2);
out = squeeze(mean(reshape([a;nan(n,2)]',size(a,2),2,[]),2))';

[skyline]

I should have been clearer. An example is a matrix

0.1  0  0.1  0;
0.1  0.1  0.1  0.1;
0.2  0       0.2     0;
0.2  0.1  0.2  0.1;
0.2  0.2  0.2  0.2;

I want to create a matrix where - the first column is the distinct element of the first column - the second column is the average of the elements of the third column of rows with the same first column. That is; where the corresponding first rows are the same

0.1 0.1;
0.2 0.1;

[Image Analyst]

It's still not clear, but here is some of it:

% Compute average of column 3 and column 1
ave = (a(:, 3) + a(:, 1))/2

[Andrei Bobrov]

A = [0.1  0  0.1  0;
0.1  0.1  0.1  0.1;
0.2  0       0.2     0;
0.2  0.1  0.2  0.1;
0.2  0.2  0.2  0.2];

[G,ii] = findgroups(A(:,1));
B = [ii,splitapply(@mean,A(:,3),G)];

Answer 2

[Image Analyst]

Here's yet another way using conv2 to use convolution to do a sliding mean of two rows:

outTemp = conv2(a, [1;1]/2, 'same'); % Moving mean
out = outTemp(1:2:end, :) % Skip every other one.