# how to find mean with nan

9 views (last 30 days)
Fani on 10 Oct 2014
Commented: Image Analyst on 10 Oct 2014
Hi!
I want to find mean with nan values and with different step. My script is:
RowDivision=[1,46,172,305,472,618,714,784,920];
nElement=diff(RowDivision); % step
meanG=cellfun(@mean,(mat2cell(G,nElement,ones(size(G,2),1)))); %mean
but if i have one only nan value i.e. from 1 to 46 the mean is nan. How can I ignore the nan? Thank you!

Azzi Abdelmalek on 10 Oct 2014
Edited: Azzi Abdelmalek on 10 Oct 2014
a=[1 2 nan 3 nan 10]
nanmean(a)
%or
mean(a(~isnan(a)))

Andrew Reibold on 10 Oct 2014
Edited: Andrew Reibold on 10 Oct 2014
Use nanmean
If you dont have it, I wrote this
For an array, x
n_nans = 0;
temp = x;
for i = 1:length(x)
if isnan(x(i))
temp(i) = 0;
n_nans = n_nans+1;
end
end
mean = sum(temp)/(length(temp)-n_nans)
Andrew Reibold on 10 Oct 2014

Image Analyst on 10 Oct 2014
If you have the Stats toolbox, use nanmeans. If you don't, you can do it in two lines of code, one to find the nans and one to get the mean:
% Create sample data
RowDivision=[1,46,172,305,472,618,714,784,920];
% Make some nans
RowDivision(2) = nan;
RowDivision(4) = nan
% Find the nans
nanElements = isnan(RowDivision)
% Compute the mean with no nans being considered.
meanNoNans = mean(RowDivision(~nanElements))
Actually you can do that in one line of code if you want.
Image Analyst on 10 Oct 2014
Oops - just like Azzi's edit which he posted while I was composing my response.