Dear R-Users, I have a model with a latent variable for a spatio-temporal process. I would like to use EM algorithm to estimate the parameters. Does anybody know how to implement the algorithm in R? Thank you very much in advance, Michela
On 04-Apr-06 michela.cameletti at unibg.it wrote:> Dear R-Users, > I have a model with a latent variable for a spatio-temporal process. > I would like to use EM algorithm to estimate the parameters. > Does anybody know how to implement the algorithm in R?Yes, many do! But the EM algorithm is a very very generic method, and the details of its implementation for a particular application depend on the formulation of the model (and the right kind of model) for the data. Also, for many models, computation of the EM algorithm (especially the "E" step) may in practice need to be done by means of Markov Chain Monte Carlo (MCMC) methods. Again, this cannot be answered without knowing more about your model. So, if you can explicitly describe you model, someone may know if such an implementation already exists in R, or people may be able to advise about how to do it. Best wishes, Ted. -------------------------------------------------------------------- E-Mail: (Ted Harding) <Ted.Harding at nessie.mcc.ac.uk> Fax-to-email: +44 (0)870 094 0861 Date: 04-Apr-06 Time: 15:40:37 ------------------------------ XFMail ------------------------------
Hi Michela, I'd like to add to Ted's message that the statistical journals represent a good 'source' for R to look at. Sometimes the authors of the papers need to implement their own algorithms and they make them available upon request. You might want to check the literature and send some emails ;) and, if you're lucky, get some R codes (that you'd probably need to adapt to your model). Some journals (e.g., statistical modelling http://stat.uibk.ac.at/SMIJ/index.html) provide software and data related to the published papers. I personally implemented an EM algorithm for mixed models and it took me time and patience. As Ted already said, if you give more details about your model you might get better answers and/or tips. Marco --- michela.cameletti at unibg.it wrote:> Dear R-Users, > I have a model with a latent variable for a > spatio-temporal process. > I would like to use EM algorithm to estimate the > parameters. > Does anybody know how to implement the algorithm in > R? > Thank you very much in advance, > Michela > > ______________________________________________ > R-help at stat.math.ethz.ch mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide! > http://www.R-project.org/posting-guide.html >