Rossana Dell'Anna
2006-Apr-06 10:21 UTC
[R] pelora extension to continuous response problem
Hi, I'm using Mark Dettling's supclust package. I have a typical numeric matrix x of explanatory variables (n cases, p features, n<< p). I have a numeric vector y of length n, containing a quantitative response variable. I am interested in grouping features in a way which is strongly predictive of response y (like in PLS approach). In paper describing Pelora it is said that Pelora can be easily adapted to this case. However it seems to me that no explicit indication is given in supclust user manual. Should I simply call aa<- pelora(x_training,y) and than predict(aa, x_test, type="probs") ? Thank You in advance rossana [[alternative HTML version deleted]]