[R] Generalized Path Seeking algorithm with R

Andrius.R andriu_kas at yahoo.com
Tue Aug 11 16:49:00 CEST 2009


Hello,

Has anybody tried to use Generalized Path Seeking algorithm introduced by
Jerome H. Friedman and implemented using R? :
http://www-stat.stanford.edu/~jhf/r-gps/GPShelp.html

I have a problem using it and I hope that you could help me. I have tried to
run the GPS model for penalized logistic Ridge and logistic Lasso
regressions on my dataset, but it seems that I am doing something wrong,
because I am getting not very good results.

I think the problem is with the definition of data for variable y. According
to GPS help: 'For classification (mode="class") variable y is a numeric
matrix with two columns respectively containing the number of each of the
two classes at the design points (rows).'

So for y matrix's second column I've tried to put 1 for actual goods and 0
for bads and for the first – inverse of the value of second column, but it
seems that it is not right. 

Does anybody have an idea how the first column has to be defined?
I would be very grateful.

Andrius

-- 
View this message in context: http://www.nabble.com/Generalized-Path-Seeking-algorithm-with-R-tp24919113p24919113.html
Sent from the R help mailing list archive at Nabble.com.




More information about the R-help mailing list