Dear all,

fminer algorithms BBRC and LAST-PM now both support the multinomial 
setting: instead of just binary target class values or numeric target 
values for regression, you may now use up to five different target 
class values.

This functionality will be exposed as a webservice within Opentox in 
the near future, in the meantime please check out the binary version, 
if your program links to libbbrc/liblast from ruby, python, java, or 
C++, or if you want to call fminer using the console.

If interested, please find details on the method and a usage example 
here: http://www.maunz.de/wordpress/opentox/2011/multinomial-fminer.
Detailed information (e.g. data representation) is also available on 
http://bbrc.maunz.de and http://last-pm.maunz.de (look for the link to 
the API documentation).

Best regards
Andreas

--
http://www.maunz.de