search for: clustindex

Hi, I'm trying to debug a function I wrote to calculate the cindex for a hierarchical tree. For this it is useful to compare my calculations with those in output from the clustindex function, in the cclust library. There's no way, however, to have the cindex value for a given output of the cclust function, as a NA value is always returned. This happens almost surely because the cindex in clustIndex is calculated only for binary data, but, in turn, I can't find a way to...

...ed, 20 Feb 2002, David Wartel wrote: > > I have to solve a clustering problem. > My first step is to determinate the number of clusters, that's why I 'm using ...snip... >... number of clusters. > A function is already implemented in R to calculate this index : > > clustIndex(cl,x, index="calinski") Where is that from? It's not part of R -- package cclust, perhaps? > where cl is the result of a clustering method , for instance: > > cclust(x,k,itermax,verbose=TRUE,method="kmeans") > > My probleme is that I can't calculate t...

...8 0.204 ... > str(x[20:30]) num [1:11] -0.315 -0.693 -0.771 0.448 0.204 ... This breaks: > cov(x) [,1] [1,] 0.9600812 > cov(x[20:30]) Erreur dans cov(x[20:30]) : fournir 'x' et 'y' ou bien 'x' en matrice And this behavior is braking function clustIndex (when used with unidimensional data), from the package cclust, file Rindexes.R, lines 137-147: ttww <- function(x, clsize, cluster) { n <- sum(clsize) k <- length(clsize) w<-0 tt <- cov(x)*n for (l in 1:k) w<- w+cov(x[...

Hi- I'd like to apply some of the cluter indices, using 'clustindex' in the cclust package, but using output from hierarchical clustering using 'hclust' in the mva package. The two packages use a different format for storing the results of clustering; has anyone coerced one into the other? Thanks! -John Barnett -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-...

> Well, that is in all good texts on R, together with the > solution: drop=FALSE. See ?"[" for the on-line details. OK. Thank you a lot. Now patched cclust and clustIndex work fine for 1D case. BTW, why not to apply the "drop=F" to these functions? I guess other users need 1D case as well. kind regards, Valery A.Khamenya --------------------------------------------------------------------------- Bioinformatics Department BioVisioN AG, Hannover

...to solve a clustering problem. My first step is to determinate the number of clusters, that's why I 'm using the Calinski index ( [tr(b)/(k-1)]/[tr(w)/(k-1)] ) which i try to maximize to have the best number of clusters. A function is already implemented in R to calculate this index : clustIndex(cl,x, index="calinski") where cl is the result of a clustering method , for instance: cclust(x,k,itermax,verbose=TRUE,method="kmeans") My probleme is that I can't calculate the Calinski's index when a cluster contains only one datapoint : Error in cov(x[cluster == l...

..., clust, cclust, mclust, multiv, normix, survey. And I didn't find any ready-to-use general purpose criterion for answering the question whether the data is "quite homogeneous" or has two or more clusters. Even for one dimension data. However, in "cclust" a "clustIndex" might be used as a raw criteria. But nothing ready to use as far as I understand. Or maybe I am wrong?! Q: are there any libraries in R with ready-to-use functions for estimation number of clusters... - ... with criterion based on entropy? - ... with criterion based...

Hi, I read a post back in April 2000 that wondered if it was possible to use hclust objects in cclust, so to take advantage of the clustIndex function to compute indexes that can be used to estimate the "true" number of clusters in the data. The answer there was that this was not possible at the moment, but that an unification of the clustering classes from different packages was probably a good idea for the future. I was won...

Is there a command to get cluster criterion for the cluster methods? SAS has its criterion, but I prefer to do it in R. If there is not a command is there code to produce criteria to choose the number of clusters? Adrian Katschke Statistics Grad Student University of Nebraska-Lincoln [[alternative HTML version deleted]]