File Exchange

image thumbnail

generalized_esd(x,a​lpha,k)

version 1.0.0.0 (1.05 KB) by Matthew Taliaferro
Applies generalized extreme Studentized deviate test for outliers at a specified confidence level

4 Downloads

Updated 12 Jul 2017

View License

Applies generalized extreme Studentized deviate test for outliers at a specified confidence level. Inputs are the population sample, significance level (suggest using 0.05) and the maximum points tested as outliers.

Comments and Ratings (1)

This is a good implementation, though the way the final check is done, does not quite work. Most of the time for multiple outliers it gives the correct answer, though with the following data, when k is set as 4 -6, 4 outliers are identified (with the fifth being borderline), and with 7 or higher setting for k, 5 outliers are identified.

I believe that it is all to do with how the ultimate check is against the last surviving gamma out of the loop as opposed to the gamma for each particular iteration.

X =
159.947697945425 164.101511262596 166.619260948399 162.190487464028 165.754224726168 166.785271327185 157.200920188827 154.686455686791 159.826481568245 161.727412902110 158.591653972205 161.818307285266 162.117381733050 157.613636210826 161.289010015366 156.836766478478 165.456166041465 159.120963501255 163.756374426894 164.863150211966 166.649101779150 169.833200000000 164.233792204613 163.592026664422 163.454633444347 162.174903826038 153.188200000000 163.211493358630 160.879373412482 164.564142001533 157.219002963402 161.485583216885 163.750617614939 163.700058042593 166.502028593753 162.973845721975 173.221900000000 161.442236968564 159.197453311462 165.535434258514

MATLAB Release Compatibility
Created with R2014a
Compatible with any release
Platform Compatibility
Windows macOS Linux