How to calculate Mean of unequal cell arrays?

I have results from 44 analyses stored in two cell arrays: 1. DRIFT (the response values) corresponding to INT (intensities). Both are 1*44 cell arrays. For example, the response from 5th analysis is stored in DRIFT{1,5} and the values are corresponding to intensity values in INT{1,5}. I stored the result in cell arrays and not as double because they are of unequal lengths. I want to find the geometric mean of all the 44 responses and plot it. However, as the responses are of different vector lengths (each analysis finish at a different point), I am unable to do so. Could someone help me?

Answer 1

[KSSV]

% make some random data for demo
N = 10 ;
A = cell(N,1) ;
for i = 1:N
    A{i} = rand(randperm(N,1),1) ;
end
%%get mean
N = numel(A) ;
% option 1: use loop
M1 = zeros(N,1) ;
for i = 1:N
    M1(i) = mean(A{i}) ;
end
% option 2: use cellfun
M2 = cellfun(@mean,A) ;

[Vishvendra Bhanu]

I believe I couldn't explain my question properly. Through this code we can get the mean of all response values stored in one column for e.g.: M(5) = mean of all values stored in DRIFT{5}. I do not want this. I want the geomean of all the values stored in the same row of all drifts - DRIFT{i}. Each DRIFT{i} is a column vector of different length. I want to find the geomean of cells corresponding to the same row number of these column vectors. Geomean() will be a column vector itself.

[Vishvendra Bhanu]

So, just to explain more: DRIFT{1,1} is a 14*1 double, DRIFT{1,2} is 17*1 double, DRIFT{1,3} is 27*1 double... and so on. I have some idea on how to find the geomean: I wish to store all these different column vectors (that is each DRIFT{1,i}) in one double matrix of size (j,44). where j is the size of the largest column vector in DRIFT. As each column is of a different size, I want to fill the blank cells with NaN and then replace with 1 (as 1 has no effect on geomean). Could you help me now?