How to average 5 rows of a vector recursively?

I have a very long 1D vector ~ 9445 long. I wish to take the average of sets of 5 down the vector and send them to a new vector for example:

1

2

3

4

5

6

7

8

9

10

would become

3

8

How can I do this easily?

Many thanks

Answer 1

[Ollie A]

V = 1:9445; % Your vector
setsize = 5;

for x = 1:length(V)/setsize
    Vnew(x) = mean(V((x-1)*setsize+1:x*setsize))
end

I hope this is simple enough.

[Guillaume]

It is much faster and simpler to reshape the vector into a matrix with 5 rows, and take the mean of the columns.

[Samuel White]

Thats great - is there a way I can get these values into a vector though please/

cheers!

[Guillaume]

It's unfortunate you've accepted that answer which is overly complicated and slow and not the matlab way. The lack of preallocation of the output is also going to be a problem if you're not careful.

However, as long as Vnew did not exist before running that code, it will produce a vector.

For the proper way to do what you want in matlab, see my answer.

[Ollie A]

I agree with Guillaume that his answer is more efficient.

Answer 2

[Guillaume]

As long as the length of your vector is a multiple of 5:

mean(reshape(yourvector, 5, []), 1)

If the length is not a multiple of 5, pad it first with NaNs to a multiple of 5 and do the same as above, with the 'omitnan' flag for mean:

paddedvector = [yourvector; nan(mod(-numel(yourvector), 5), 1)];  %pad to a length multiple of 5
mean(reshape(paddedvector, 5, []), 1, 'omitnan')