[BioC] (no subject)
dante2442 at yahoo.it
Sun Jul 18 00:48:29 CEST 2004
i followed for long time (about one year) the messages of the sam-problem about the study of the microarray data in this mailing list.
i'm statistician afferent of the department of statistic of the univerisity of padua - italy and i 'm developing any studyes about microarray.
in particular i concentrate about milan-materials (may 2003) which it can possible to download under the site http://www.bioconductor.org
my work try to find any alternative estimators of lamda about the penalized logistic regression.
the milan-materials proposed only AIC (akaike information criterion) for the estimate of the lamda parameter.
i proposed , during my resarches the BIC criterion and a AIC-modified which uses the pseudo-likelyhood. but it's not all.
i developed an algorithm that works under the free software R and it estimate the penalized parameter lamda whith different criterion:
c.. AIC-modified (about pseudo-likelyhood)
the previous algorithm that i downloaded from www.bioconductor.org doesn't work very well.
so, i'd like to signal that now the new algorithm works well.
do you guess it's correct to signal these problems here in the mailing?
do you think anybody could be interest?
a copy of this work can be found under the following url:
http://www.microarray.altervista.org (italian language but early in english)
Dipartimento di Scienze Statistiche, Universita' di Padova
Via C. Battisti 241-243, 35121 Padova (Italy)
email dante2442 at yahoo.it
[[alternative HTML version deleted]]
More information about the Bioconductor