Prev: display figures in matlab distributed computer server
Next: colormap help
From: James on 12 Apr 2010 11:36
Here is what i have so far:
A = [123; 43; 89; 100; 43; 67; 98; 43;];
B = [12; 9; 8; 2; 33; 6; 5; 18;];
C = [A,B]

The output is thus,
C =
123 12
43 9
89 8
100 2
43 33
67 6
98 5
43 18

I want to write a script that reads through A to see if there are any repeated numbers, and calculates the average B value for any repeated numbers. Then, the script automatically replace every B value with that average value.
For example, There are 3x43 which have corresponding B values of 9, 33, and 18. Therefore, the average for 43 would be 20 ((9 + 33 +18)/3). I would love the script to replace 9, 33, and 18 with 20.
As a MATLAB newbie, I have no idea where to start. Please help.
Any kind of help offered would be greatly appreciated.
From: Matt Fig on 12 Apr 2010 13:13
Depending on the sizes of, and value ranges in A and B, your fastest method might vary. Here is one vectorized method that seems to be pretty quick for larger arrays with larger values.


% Data
A = [123; 43; 89; 100; 43; 67; 98; 43];
B = [12; 9; 8; 2; 33; 6; 5; 18];


% Engine
[As,idx] = sort(A);
As = cumsum([1;diff(As)~=0]);
C = accumarray(As,B(idx),[],@(x) sum(x)/length(x));
[idx,idx] = sort(idx);
C = [A,C(As(idx))];
From: James on 13 Apr 2010 04:15
"Matt Fig" <spamanon(a)yahoo.com> wrote in message <hpvkb1$5gt$1(a)fred.mathworks.com>...
> Depending on the sizes of, and value ranges in A and B, your fastest method might vary. Here is one vectorized method that seems to be pretty quick for larger arrays with larger values.
>
>
> % Data
> A = [123; 43; 89; 100; 43; 67; 98; 43];
> B = [12; 9; 8; 2; 33; 6; 5; 18];
>
>
> % Engine
> [As,idx] = sort(A);
> As = cumsum([1;diff(As)~=0]);
> C = accumarray(As,B(idx),[],@(x) sum(x)/length(x));
> [idx,idx] = sort(idx);
> C = [A,C(As(idx))];

Hey Matt, It's working 100%.
What I'm trying to do now is to use the function you showed the other day.
Here is what i have after adding your script and the interp1 function
A = [123; 43; 89; 100; 43; 67; 98; 43];
B = [12; 9; 8; 2; 33; 6; 5; 18];
% Engine
[As,idx] = sort(A);
As = cumsum([1;diff(As)~=0]);
C = accumarray(As,B(idx),[],@(x) sum(x)/length(x));
[idx,idx] = sort(idx);
C = C(As(idx));
f = @(x) interp1(A,C,x);
f(43)
This is what i get:
??? Error using ==> interp1 at 261
The values of X should be distinct.

Error in ==> @(x)interp1(A,C,x)

Why isn't this working?
I thought assigning the same B value for repeated A should get the script running.
From: James on 13 Apr 2010 10:05
Anyone?
From: Bruno Luong on 13 Apr 2010 10:24
"James " <cche5398(a)uni.sydney.edu.au> wrote in message <hq1tn9$13s$1(a)fred.mathworks.com>...
> Anyone?

If your goal is to interpolate, then you have to remove repeated values. John's consolidator is the code for that. http://www.mathworks.com/matlabcentral/fileexchange/?term=consolidator

Or directly do:

[Au I J] = unique(A);
C = accumarray(J(:), B(:)) ./ accumarray(J(:), 1); % compute the mean
f = @(x) interp1(Au,C,x);
f(43)

% Bruno
 |  Next  |  Last
Pages: 1 2 3
Prev: display figures in matlab distributed computer server
Next: colormap help