[R] Outlier detection with the dprep library

[Talbot Katz]

Hi.

The dprep library has at least three different methods for outlier 
detection: baysout, mahaout, robout.

I wanted to test them on a very simple data set:

vrmat<-cbind((1:22),c(8,14,14,17,21,20,27,23,25,33,31,32,30,36,37,40,42,44,52,61,81,265))

As you can see by eyeballing this, the last point is a very good outlier 
candidate, and maybe the second to last point, too.  Anyway, I couldn't get 
a single one of the dprep methods to work.  They all gave me 
incomprehensible error messages (incomprehensible to me, at least).


>robout(vrmat,1)
Error in matrix(0, nrow, rep) : non-numeric matrix extent
>

>mahaout(vrmat,1)
Error in cov.rob(tempo, method = "classical") :
        at least 2 cases are needed
>


>baysout(vrmat)
Error in as.matrix(out.sort[1:n, ]) : subscript out of bounds
>


Can anybody tell me what I'm doing wrong?  I know there are a bunch of 
outlier detectors in R, can anyone recommend one (and perhaps show an 
example, preferably with my vrmat data set)?

Thanks!

--  TMK  --
212-460-5430	home
917-656-5351	cell