How to aggregate data based on values in a vector?
4 views (last 30 days)
Show older comments
HeadInTheClouds
on 14 Dec 2017
Edited: Lateef Adewale Kareem
on 7 Apr 2019
I have matrix A containing 128 columns and m rows. I want to aggregate columns in matrix A based on values in a vector B containing n columns. In the instance that:
A = rand(1000,128);
B = [32 45 58 62 73 84 103 112];
I would want to sum columns 1:32 33:45 46:58 59:62 63:73 74:84 85:103 104:112 and 113:128 of matrix A to create a new 1000x9 matrix C. Right now, my code to perform this process is:
C = horzcat(...
sum(A(:,1:B(1)),2),...
sum(A(:,B(1)+1:B(2)),2),...
sum(A(:,B(2)+1:B(3)),2),...
sum(A(:,B(3)+1:B(4)),2),...
sum(A(:,B(4)+1:B(5)),2),...
sum(A(:,B(5)+1:B(6)),2),...
sum(A(:,B(6)+1:B(7)),2),...
sum(A(:,B(7)+1:B(8)),2),...
sum(A(:,B(8)+1:end),2));
I would like to automate this process so that I can change the number of columns in vector B without altering my aggregation code
Many thanks!
0 Comments
Accepted Answer
Jos (10584)
on 14 Dec 2017
Edited: Jos (10584)
on 14 Dec 2017
Here is one flexible approach:
A = cumsum(ones(3,10),2) % a simple example, easy to check
B = [2 5 6]
B2 = [0 B size(A,2)]
C = arrayfun(@(k) sum(A(:,(B2(k)+1):B2(k+1)),2), 1:numel(B2)-1,'un',0)
C = cat(2,C{:})
2 Comments
Lateef Adewale Kareem
on 7 Apr 2019
Edited: Lateef Adewale Kareem
on 7 Apr 2019
Another way to look at it is to define starting and ending B(s)
B1 = [1,B+1]; B2 = [B,size(A,2)];
C = arrayfun(@(k1,k2)sum(A(:,k1:k2),2),B1,B2,'un',0);
C = cat(2,C{:})
More Answers (0)
See Also
Categories
Find more on Portfolio Optimization and Asset Allocation in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!