[R] Randomness tests

Johannes Huesing johannes at huesing.name
Thu Sep 24 08:04:40 CEST 2015



Giorgio Garziano <giorgio.garziano at ericsson.com> [Tue, Sep 22, 2015 at 10:24:25PM CEST]:
>Hi,
>
>to test randomness of time series whose values can only be +1 and -1, are all following
>randomness tests applicable or only a part of ?

I don't know the details of all those tests but the general problem is
that the alternative hypothesis of "non-randomness" is of infinite
dimension and thus not testable. It depends on which deviations from
randomness you are interested in. You may also want to look into 
scan statistics but I presume that difference.sign.test() will be 
sensitive in a similar direction.

-- 
Johannes Hüsing               
http://derwisch.wikidot.com



More information about the R-help mailing list