different z values for one x y coordinate

hei,

my problem is, that i have xyz points and it happens, that for one identical x,y value i get several z values.

now i want to get all z values for one x,y koordinate.

i tried first with unique so i now all x,y koordiates. but now i would like to get all z values which belong to this x,y.

thanks for help. ;-) markus

Answer 1

[Markus]

%%looking for unique xy values

      uniCoord=unique(data(:,1:2),'rows');
      uniCoord(:,3:12)=NaN;

%%writing for each xy value in column
% 3-9 all corresponding z values
% 10 amount of corresponding z values
% 11 median of z values
% 12 mean of z values

    for i =1:size(uniCoord,1)
        A=(data(:,1)==uniCoord(i,1)&data(:,2)==uniCoord(i,2));
        B=data(A,3)';
        uniCoord(i,3:2+size(B,2))=B;

        uniCoord(i,10)=size(B,2);
        uniCoord(i,11)=median(uniCoord(i,3:2+size(B,2)));
        uniCoord(i,12)=mean(uniCoord(i,3:2+size(B,2)));
    end

Answer 2

[Sean de Wolski]

You could use the index output arguments of unique along with accumarray(). What do you want to do with multiple z values?

doc unique %look at 3rd output
doc accumarray

More per comments:

xy = [1 2; 1 2; 1 2; 3 4; 5 6; 5 6; 7 8]; %xy pairs
z = (1:length(xy))';

[~,idxk,idxn] = unique(xy,'rows'); %idxk = kept rows, idxn = subs_rows
zmean = accumarray(idxn,z,[],@mean); %get the mean of the groups
xyz = [xy(idxk,:) zmean] %concatenate

[Markus]

the idea is to average all z values at the same geographical xy- position.

[Sean de Wolski]

yes, accumarray with unique and mean() would be ideal. Though a for-loop works well too!

[Markus]

i checked the doc for accumarray, but i dont understand how to use it in this case. could you give an example for this case ..

the loop needs a lot of calculating power .. we have in the smallest datasets 4,500,000 rows :)

[Sean de Wolski]

Sure. I'll never turn down any opportunity to use accumarray (or bsxfun)!

[Markus]

ok. thats a nice solution thanks. :)