# Average values for duplicates in (:,1)

2 views (last 30 days)
aaron Harvey on 12 Oct 2015
Answered: Andrei Bobrov on 12 Oct 2015
Hi, I have written a script with a for loop in it that plots two measurements (water tempurature and salinity) for a given depth and a given x position. There are only 7 different x locations. I have a matix B being produced in a for loop for various depths, a new B for each new depth. Where B=[xdistance,Temp,Sal] e.g
B=
000 14 33
000 14 34
700 13 35
700 13 36
700 12 35 . . .
how can I average B such that there are no duplicates in B(:,1)
B=
000 14 33.5
700 12.67 35.67 ...
without having to specify the length of B as it will change for each depth (each iteration of the for loop).

Andrei Bobrov on 12 Oct 2015
[A,~,ii] = unique(B(:,1),'stable');
x = B(:,[2,3]);
[a,b] = ndgrid(ii,1:2);
A(:,2:3) = accumarray([a(:),b(:)],x(:))./(accumarray(ii,1)*ones(1,size(x,2)));

Stephen23 on 12 Oct 2015
Edited: Stephen23 on 12 Oct 2015
You can use unique and accumarray very efficiently:
B = [...
000 14 33
000 14 34
700 13 35
700 13 36
700 12 35]
[A,~,z] = unique(B(:,1),'stable');
A(:,2) = accumarray(z,B(:,2),[],@mean);
A(:,3) = accumarray(z,B(:,3),[],@mean);
Generates this output matrix:
>> A
A =
0.00000 14.00000 33.50000
700.00000 12.66667 35.33333