# How can one "summarize a matrix with the first four columns"?

1 view (last 30 days)
alpedhuez on 29 Jul 2020
Commented: Jon on 7 Aug 2020
First, I have a matrix like
1 2 3 4 1.1
1 2 3 4 3
1 2 3 5 4
2 3 4 1 6
...
Second, what I would like to do is to summarize the matrix with the first four columns. That is, here, I want to take an average of the entries of the fifth item for the rows that has the same first four columns. In this example, it will be
1 2 3 4 2.05
1 2 3 5 4
2 3 4 1 6
...

Jon on 29 Jul 2020
Edited: Jon on 29 Jul 2020
Here is one way to do it. There might be some clever way to fully vectorize (eliminate the loop) this, but this will work if performance on huge arrays isn't an issue
A = [1 2 3 4 1.1;1 2 3 4 3; 1 2 3 5 4; 2 3 4 1 6]
% find rows with unique first four elements
[C,~,ic] = unique(A(:,1:4),'rows')
% summarize by finding average of elements in fifth column over rows with
% same first four columns
for k = 1:length(ia)
% use logical indexing to find rows with unique first four elements
C(k,5) = mean(A(ic==k,5))
end

alpedhuez on 29 Jul 2020
Thank you. There might be "accumarray" solution in the third part that I would like to understand.
Jon on 7 Aug 2020
Good idea using accumarray. I didn't know that command. Definitely looks like it has some possibilities.
C(:,5) = accumarray(ic,A(:,5))./accumarray(ic,1)
instead of the for loop works for this case. I would have to think about it more to know if there are any edge cases where this would fail, but I think it maybe a nice way to do it.