[Traminer-users] Nominal autocorrelation

so.bellit so.bellit at laposte.net
Mon Apr 30 10:50:20 CEST 2012


 
Bonjour,

  J'ai effectué une classification ascendante hiérarchique à partir de Traminer. J'ai utilisé la fonction seqrep pour retrouver le médoid de chaque classe. De plus, par la fonction print (medoid,...), je peux aussi récupérer le numéro de ligne correspondant à la trajectoire représentative. Cependant, lorsque j'utilise la fonction seqrplot qui permet d'obtenir plusieurs séquences représentatives, je ne parviens pas à récupérer le numéro de ligne ou tout simpement l'dentifiant auquel elles correspondent. Auriez vous une idée là dessus?

Cordialement
Sonia



> Message du 30/04/12 09:34
> De : "Gilbert Ritschard"
> A : "Users questions"
> Copie à :
> Objet : Re: [Traminer-users] Nominal autocorrelation
>
> Hi Peter,
>
> Using the seqformat() function to transform your sequences into SRS
> (shifted-replicated sequence) form (see Ritschard & al., 2009), it is
> easy to compute some nominal auto-association of any order. Here is an
> example where we compute an auto-cv of order 1 (cv being Cramer's v).
>
> libarary(TraMineR)
> data(biofam)
> biofam.srs <- seqformat(biofam[,10:25], from="STS", to="SRS")
> nc <- ncol(biofam.srs)
> x1 <- table(biofam.srs[,nc],biofam.srs[,nc-1])
> chisq <- chisq.test(x1, correct=FALSE)$statistic
> names(chisq) <- ""
> cv1 <- sqrt(chisq/(sum(x1)*min(dim(x1)-1)))
> print(cv1)
>
> Reference for the SRS format:
>
> Ritschard, G., Gabadinho, A., Studer, M. & Müller, N.S. (2009),
> "Converting between various sequence representations", In Ras, Z. &
> Dardzinska, A. (eds) /Advances in Data Management/. Series: /Studies in
> Computational Intelligence/. Volume 223, pp. 155-175. Berlin: Springer.
> (preprint available on the TraMineR page)
>
> Cheer.
> Gilbert
>
> On 24-Apr-12 21:15, Peter Flom wrote:
> >
> > Good afternoon
> >
> > I just got TraMineR – what a great package! It’s going to save me a
> > lot of time and effort.
> >
> > I have a data set with a bunch of sequences of nominal states (for
> > now, anyway, 80 sequences, with each having one of two states, but it
> > may also be necessary to do this with 4 states and more sequences)
> >
> > My client wants some nominal equivalent of autocorrelation. Looking
> > through the documentation, the closest thing seems to be turbulence,
> > or perhaps inverse of turbulence, but that’s per sequence, not per
> > data set. Of course, I could find mean or median turbulence, or mean
> > or median inverted turbulence, but I wondered if there is some already
> > developed measure for what I want.
> >
> > Thanks in advance
> >
> > Peter
> >
> > Peter Flom
> >
> > Peter Flom Consulting
> >
> > http://www.statisticalanalysisconsulting.com/
> >
> > http://www.IAmLearningDisabled.com
> >
> >
> >
> > _______________________________________________
> > 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
>
> --
> Gilbert Ritschard, Department of Economics and
> Institute for Demographic and Life Course Studies,
> University of Geneva, 40, bd du Pont-d'Arve, CH-1211 Genève 4, Switzerland
> http://mephisto.unige.ch
>
> _______________________________________________
> 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
> 

Une messagerie gratuite, garantie à vie et des services en plus, ça vous tente ?
Je crée ma boîte mail www.laposte.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/traminer-users/attachments/20120430/dfeea723/attachment.html>


More information about the Traminer-users mailing list