Hi all: Melissa C. here at Princeton told me she has some C versions of the code that we could use, and also mentioned some matlab versions, which I subsequently found on the web.
As such, it should be pretty easy to implement both LARS and ENET right in PyMVPA, like we did for SMLR. I'll let folks know when I've made progress on this... Best, P On 2/21/09, Michael Hanke <[email protected]> wrote: > Hi all, > > just for the record: Both LARS and ENET classifiers will not work on > Debian etch, since the corresponding RPy prevents 'dim' being reported > for NumPy array, i.e.: > > >>> print rpy.r.dim(N.ones((2,5))) > None > > That breaks both classifiers with: > > ====================================================================== > ERROR: testLARSSensitivities (__main__.LARSTests) > ---------------------------------------------------------------------- > Traceback (most recent call last): > File "mvpa/tests/test_lars.py", line 60, in testLARSSensitivities > clf.train(data) > File "/home/hanke/hacking/pymvpa/mvpa/clfs/base.py", line 354, in train > result = self._train(dataset) > File "/home/hanke/hacking/pymvpa/mvpa/clfs/lars.py", line 149, in _train > use_Gram=self.__use_Gram) > RException: Error in rep(1, n) : invalid 'times' argument > > > > This is probably close to a non-issue since etch is now 'oldstable', but > somebody might still stumble upon it. > > > Michael > > > -- > GPG key: 1024D/3144BE0F Michael Hanke > http://apsy.gse.uni-magdeburg.de/hanke > ICQ: 48230050 > > _______________________________________________ > Pkg-ExpPsy-PyMVPA mailing list > [email protected] > http://lists.alioth.debian.org/mailman/listinfo/pkg-exppsy-pymvpa > _______________________________________________ Pkg-ExpPsy-PyMVPA mailing list [email protected] http://lists.alioth.debian.org/mailman/listinfo/pkg-exppsy-pymvpa
