| To: | s-news@lists.biostat.wustl.edu |
|---|---|
| Subject: | multivariate tree |
| From: | Erik van Zwet <evanzwet@math.leidenuniv.nl> |
| Date: | Thu, 31 Jul 2003 14:09:49 +0200 (MEST) |
Dear all, I'd like to use Splus or R to fit a regression tree to a multivariate response variable. In fact, the response variable is a function sampled at 96 time points. I think the deviance at a node should simply be the L2 distance of all the members to their mean vector. Hopefully, someone has some code that does this. If not, then it seems rpart() could be modified, but having no C experience at all, I'd need some guidelines or examples on how to do that. Thanks, Erik van Zwet |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Variance dispersion graphs, Coetzer, Roelof (RLJ) |
|---|---|
| Next by Date: | coefficients for a low pass binomial filter, Stefano Sofia |
| Previous by Thread: | Variance dispersion graphs, Coetzer, Roelof (RLJ) |
| Next by Thread: | coefficients for a low pass binomial filter, Stefano Sofia |
| Indexes: | [Date] [Thread] [Top] [All Lists] |