Function similar to MS Excel "subtotal"?

I have a 150x2 matrix in which the first column contains numbers that can be considered 'grouping variables' and the second column contains values associated with those grouping variables. So a small 5x2 version would look like this

1 23
1 29
1 43
2 38
2 22

I'd like to average the values associated with each grouping variable. So the above example would yield an average of 31.67 (for variable 1) and 30 (for variable 2). In Excel this can be achieved using Data --> "Subtotals". Does anybody know of a similar function or method to achieve this in Matlab?

Thanks

Answer 1

[Titus Edelhofer]

Hi George,

I don't know a single function, but the functionality is not difficult to achieve:

x = [1 23; 1 29; 1 43; 2 38; 2 22]

% find the different indices:
id = unique(x(:,1));
subTotals = zeros(size(id));
for i=1:numel(id)
  subTotals(i) = mean(x(id==id(i), 2));
end

Titus

Answer 2

[George]

Hi Titus, Thanks for the quick reply. The code doesn't quite work because 'id' needs to be the indices of the unique values. Your code got me 95% there though, so with slight modification I've got it working.

x = [1 23; 1 29; 1 43; 2 38; 2 22] 
% find the different indices: 
[vals vec id] = unique(x(:,1)); 
subTotals = zeros(size(vals)); 
for i=1:numel(id) 
  subTotals(i) = mean(x(id==id(i), 2));
end

Thanks again!

[Titus Edelhofer]

Hi George,

yes of course! My code accidently worked because for the example vals and id are the same ...

Titus

Answer 3

[Kelly Kearney]

If your grouping variable includes only integers, then

x = [...
1 23
1 29
1 43
2 38
2 22];

subtot = accumarray(x(:,1), x(:,2), [], @mean);