Asked by Fly Haynes
on 21 Mar 2014

Hi, I can't seem to get around how to do this. I have 12xm matrix (a1) of which I want to calculate the linear rank correlation rho an pvalues (using Corr function and pearson correlation type) from paired columns vectors, i.e. between 6 rows in each column and the next 6 rows in that column (these are replicate measurements, designated rep1 and rep2, respectively), and I want to iterate that with a for loop for all the m columns in the matrix a1.

I have it working so that it spits out the rho values in a 1xm vector, but I don't know how to get the script to do that also for the p-values associated with each correlation.

this is the script I have:

results=zeros(1,size(a1,2)); rep1=a1(1:6,1:size(a1,2)); rep2=a1(7:12,1:size(a1,2));

for repit = 1:size(a1,2); results(repit) = corr(rep1(:,repit),rep2(:,repit)); end

I know that [r,p] = corr(x,y) gives me both the r and p-value, but how do I build this into the iterative loop that I have going so that it does that for each each two column vectors in every columns of the a1 matrix.....?

I hope I am making this clear enough... otherwise, don't bother asking for clarifications.

Thank you.

*No products are associated with this question.*

Answer by Mischa Kim
on 21 Mar 2014

Edited by Mischa Kim
on 21 Mar 2014

Accepted answer

Fly, you could use a `cell` array:

... [r{repit} p{repit}] = corr(rep1(:,repit),rep2(:,repit)); ...

The results from the first iteration are then accessed via `r{1}` and `p{1}`, from the second in `r{2}`...

If the r- and p-values are scalars, you can also just use simple numeric arrays:

[r(repit) p(repit)] = corr(rep1(:,repit),rep2(:,repit));

The advantage of cell arrays for non-scalars is that you don't need to worry about matrix dimensions.

Opportunities for recent engineering grads.

## 0 Comments