moving window chi square test for outlier removal

Hi,

I'm trying to apply a chi-square test with a moving window 10 on a vector. I want ot replace the outliers with NaN within that window, if chi-sqaure >2*standard deviation. Any suggestions?

n=length(yy)
R(i) = abs( yy(i) - mean(yy) ) / std(yy);
% statistical test to find outliers
lamda = 2*sd(yy)
if R(i) > lambda
    index = is(n-i+1:end);   
    index = [ index1 is(n-i+1) ];
end

Answer 1

[Cris LaPierre]

One thought would be to use the interactive task in live scripts to find, fill or remove outliers. It might not be sophisticated enough to fully do what you want, but it definitely can get you started. Once you have what you want, you can export it as code and further modify that to meet your needs.

[nlm]

Thanks for the suggestion. However, It has limited options and does not have chi square test for outlier detection. Do you have any ideas for that ?