# Getting the mean value of a row in an array?

1 view (last 30 days)
Chris Lambrecht on 17 Sep 2015
Edited: Jon on 18 Sep 2015
I have the code
filename='Laramie2005_2015.dat'; %Wind data file
iminsamp=53;
yearno = 2006:2009;
for mo = 1:5
for yearidx = 1 : length(yearno)
yr = yearno(yearidx);
[dttm{mo,yearidx}, timemin{mo,yearidx}, wnddatenum{mo,yearidx}, wndspeed{mo,yearidx}, wnddir{mo,yearidx}, pres{mo,yearidx}, temp{mo,yearidx}] = RdNCDCData(filename, mo, yr, iminsamp);
end
end
which gives my desired variables as an array that is [mo,yearidx] in size with each one of the values varying in size as a 1:n row maxtix. I am looking to get the average for each row but am only getting the average of each column for each row using ave(x)=arrayfun(@(x) mean(wndspeed{x}),x) which in turn produces a matrix. Is there a way to get a single average for each month using something like that?
Kirby Fears on 17 Sep 2015
Chris,
I'm having trouble understanding your question. Do want to take the average (collapsing dimension 2) of an N x M cell where each cell contains a 1 x L numeric array (where L varies for each cell)? Is the desired output a N x 1 vector of averages?

Jon on 17 Sep 2015
Edited: Jon on 18 Sep 2015
If you have an array called myarray that is mo x yearidx (as you state in your question) and you want the average of all mo == j, then you could just write
meanmo(j) = mean(myarray(myarray(:,1)==j),:);
where j is the index of the month you want to average over. Just replace "myarray" with whichever variable you want to average over.
Edit: Kirby pointed out that you're using cell storage instead of matrix, so to use my method you'd need to convert your cell to a matrix first:
myarray = cell2mat(temp);
Kirby Fears on 17 Sep 2015
Jon,
In Chris' example, he appears to have a "mo x yearidx" cell, not an array. He's using curly brackets inside of his for loop.

Kirby Fears on 17 Sep 2015
Edited: Kirby Fears on 17 Sep 2015
Chris,
I think you're looking for one of two things, but I'm not sure which. Try these out and see if it's the output you're looking for. If not, please try to clarify your question.
** Creating example data
sample{1,1}=ones(1,5);
sample{1,2}=2*ones(1,6);
sample{2,1}=3*ones(1,4);
sample{2,2}=4*ones(1,5);
1. First interpretation of question: one mean per cell.
means=cellfun(@(c)mean(c),sample);
2. Second interpretation of question: one mean per "mo".
sums=cellfun(@(c)sum(c),sample);
counts=cellfun(@(c)numel(c),sample);
means=sum(sums,2)./sum(counts,2);
Hope this helps.