How to summarize a matrix based on unique groups

I have a vector A=[1 2 1 4; 5 3 5 7; 6 1 8 9], the first row represent age group, and the second row represents skill level, and the third row is the count of people. From A, I want to produce matrix B where is the compact form of A, in other words, if two or more columns represent the people in the same age and skill level group, we remove all of them and only keep one of them. For our example, A([1 2],1) = A([1 2],3) = [1;5], therefore B=[1 2 4; 5 3 7; 14 1 9].

Answer 1

[Star Strider]

Try this:

A=[1 2 1 4; 5 3 5 7; 6 1 8 9];

[Au, ~, ic] = unique(A([1 2],:)', 'rows');          % Unique Columns
R3 = accumarray(ic, A(3,:));                        % Sum Similar Elements In Row #3
B = [Au R3]'                                        % Desired Result

B =
     1     2     4
     5     3     7
    14     1     9

Answer 2

[Kazi]

Thank you! This is the right answer. But, it is slow. I ran the profiler and it looks like unique function sorts the data and it takes a significant amount of time. Is there any alternative faster solution?

[Star Strider]

My pleasure!

‘I ran the profiler and it looks like unique function sorts the data and it takes a significant amount of time. Is there any alternative faster solution?’

Not that I am aware of. The unique() function is necessary for the code to work, since it detects the columns with the same values in the first two rows of ‘A’.