How do you replace NaN values in column n with the average of the all the values in column n?

I have a 9172x27 matrix called ProjectData_E where there is a handful of missing data. I have found a way to find the average value for each column, n, while passing over the NaN values and that are stored in a 1x27 matrix called column_mean. I have tried playing around with the isnan function and creating a loop, but I am struggeling on how to replace NaN values in column 1 with the average of all the values of column 1 and so on. Is there another way to go about this?

for j = 1:27

ProjectData_E(isnan(ProjectData_E)) = column_mean(:,j);

end

I have tried playing around with a simple loop like this, but either it does not work at all or all NaN values are replaced with the mean of column 1.

Answer 1

[David Hill]

r=rand(9172,27);
r(randi(247644,1,10000))=nan;%add some nan's
m=mean(r,'omitnan');
[~,idx]=find(isnan(r));
r(isnan(r))=m(idx);