Am 08.02.2015 20:24 schrieb "Neil Bowers" <neil.bow...@cogendo.com>:
> I want to make a few modules that each do a specific type of statistical > detection of outliers in a data set. They may or may not use PDL. > > [….] > > I'm thinking good names would be: > - Math::Stats::Outlier::GrubbsTest > - Math::Stats::Outlier::ESDTest > - Math::Stats::Outlier::TietjenMooreTest > > Any suggestions? > > I would not care publicly about the names of the three tests, just about the expected number of outliers in the result. So I would go with one module only: *Statistics::Outliers* I.e Grubbs with exact one, Tietjen-Moore with exact n, and the slowest ESD with up to n outliers. So Grubbs + Tietjen-Moore can be combined by looking the n argument, and ESD can be a seperate function. > > Most statistics modules seem to live in *Statistics::*, so > *Statistics::Outlier::* might be a better home. > > Other thoughts on namespace: > > *Statistics::Outliers::* > *Statistics::OutlierDetection::* > * Statistics::OutlierTest::* > > With the latter you could then have *Statistics::OutlierTest::Grubbs* > > Neil > >
