Many thanks,
 Anca

Murray Jorgensen <[EMAIL PROTECTED]> wrote: I have not used Latent Gold but I 
can talk in general about the use of 
mixture models in clustering. The statement "based on means and not on 
probabilities" seems wrong.

All mixture models are based on viewing the data as a sample from a 
finite mixture of probability distributions, each with an assumed 
parametric form. When the mixture components are well=separated in the 
data the clustering obtained is reasonably robust to the assumed form. 
When the components overlap substantially there can be identifiability 
problems and the form of the assumed distributions becomes more 
influential on the solution.

Traditional statistical tests based on asymptotics are not of much 
relevance because the number of parameters for these models is usually 
large in comparison to the number of observations. Resampling may be of 
use though, not so much for obtaining confidence regions for component 
model parameters and mixture proportions as for see whether the main 
features of a maximum likelihood solution are reproduced in resamples.

John Ubersax's Latent Class Pages
http://ourworld.compuserve.com/homepages/jsuebersax/

are a good resource for mixture models, although most attention is given 
to discrete variables.

David Dowe's mixture modelling page
http://www.csse.monash.edu.au/~dld/cluster.html

is a useful resource and collection of links.

Murray Jorgensen

SUBSCRIBE CLASS-L Anonymous" wrote:
> Hello
> 
> Does anyone know why p-values and chi-squared statistics are not available in 
> Latent Gold summary output for models using continuous variables and what is 
> the statistical explanation behind it?
> 
> Also, how reliable is the classification with continuous variables in latent 
> gold 
> given the fact that it is based - from my understanding - on means and not on 
> probabilities?
> 
> Many thanks
> Anca
> 
> ----------------------------------------------
> CLASS-L list.
> Instructions: http://www.classification-society.org/csna/lists.html#class-l
> 
> 

-- 
Dr Murray Jorgensen      http://www.stats.waikato.ac.nz/Staff/maj.html
Department of Statistics, University of Waikato, Hamilton, New Zealand
Email: [EMAIL PROTECTED]                                Fax 7 838 4155
Phone  +64 7 838 4773 wk    Home +64 7 825 0441    Mobile 021 1395 862

----------------------------------------------
CLASS-L list.
Instructions: http://www.classification-society.org/csna/lists.html#class-l


 
---------------------------------
Everyone is raving about the all-new Yahoo! Mail beta.

----------------------------------------------
CLASS-L list.
Instructions: http://www.classification-society.org/csna/lists.html#class-l

Reply via email to