[Traminer-users] seqrep and distance matrix

Alexis Gabadinho Alexis.Gabadinho at unige.ch
Thu Jun 10 17:14:28 CEST 2010


Hi Laurent,

Here is the answer to your first question. "seqrep" works fine with 
"half" matrices produced with the full.matrix=FALSE option. However, the 
plot function, "seqrplot", does not work. This is fixed (thanks to your 
mail). The feature will be available as of tomorrow in the development 
version of TraMiner which you can install by following these instructions:

http://mephisto.unige.ch/traminer/install.shtml#devel

Nevertheless, if you don't want grouped plots you can plot the 
representative sequences with the current (1.6) version, by using the 
plot() method:

mvad.om <- seqdist(mvad.seq, method="OM", sm="TRATE", full.matrix=FALSE)
mvad.rep <- seqrep(mvad.seq, dist.matrix=mvad.om)
plot(mvad.rep)

Notice also that there is a minor bug in the seqrplot function 
concerning the displayed coverage statistics if you use another 
neighborhood diameter than the default 0.10. This is fixed in the 
development version as well.

Best,
Alexis.


Laurent Lesnard a écrit :
> Dear TraMineR makers and users
>
> Is it correct that seqrep does not accept distance matrix produced by 
> seqdist with the option full.matrix=FALSE? If so, what to you suggest 
> to circumvent this problem: rerun the OM analysis with full.matrix=TRUE?
>
> Another unrelated question: my experience is that TraMineR is very 
> fast but cluster analysis is not. Any plan to introduce mixed 
> clustering method (partitioning then ascending cluster analysis)?
>
> Thanks a lot
>
> Laurent
> _______________________________________________
> Traminer-users mailing list
> Traminer-users at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/traminer-users 
>



More information about the Traminer-users mailing list