I have a 5X126 matrix that I am splitting into a [5,21,6] matrix using reshape. I want to average the values of neighboring columns in each of these submatrices and store them into a final 3-d matrix that will be [5 20 6]. How can i write a for loop to do this?
I am very new to Matlab so any help would be appreciated. Thanks!

3 Comments

KSSV
KSSV on 4 Mar 2020
What average/ mean you want? Check the dimensions of the output once.
RefBlock = reshape(FinalRef,5,21,m./21); %%breaking down matrix into separate blocks
SamBlock = reshape(FinalSam,5,20,n./20);
RefBloAvg = zeros(5,20,m./21);
for i = 1:20
for j = 1:(m./21)
RefBloAvg(:,i,j) = (FinalRef(:,i,j) + FinalRef(:,i+1,j))./2;
end
end
this is what I wrote to average neighboring values in my new sub-matrix but I get the error "Index in position 3 exceeds array bounds (must not exceed 1)."
"I want to average the values of neighboring columns" is a little ambiguous, if we consider a smaller example, such as the following 5 by 5 matrix
>> M = randi(10,[5 5])
M =
8 9 2 4 6
1 8 1 5 9
9 6 10 7 4
10 2 4 1 5
10 4 3 9 1
Is this the output you're expecting on each page as follows:
>> result = M(:,1:end-1) + diff(M,1,2)./2
result =
8.5000 5.5000 3.0000 5.0000
4.5000 4.5000 3.0000 7.0000
7.5000 8.0000 8.5000 5.5000
6.0000 3.0000 2.5000 3.0000
7.0000 3.5000 6.0000 5.0000

Sign in to comment.

 Accepted Answer

Assuming that for the average of columns you want the result that Turlough specified, then
RefBloAvg = movmean(RefBlock,[0 1],2);
RefBloAvg(:,end,:) = [];
Note that this does not need to be in a loop of any kind. It works on the entire array at once.

More Answers (2)

Here's the method I was suggesting. The idea is to take the first value and add half the difference (effectively interpolating)
RefBloAvg = RefBlock(:,1:end-1,:) + diff(RefBlock,1,2)./2;
Matt J
Matt J on 4 Mar 2020
Edited: Matt J on 4 Mar 2020
Another guess,
result=convn(RefBlock,[1,1]/2,'valid');

Categories

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!