I had a cross-validation script for LogisticRegression completely break and after seeing this thread I suspect that scale_C is at fault.
Did the default behavior change from not scaling C to scale_C = True between version 0.9 and the current git bleeding edge? Updating from 0.9 to the git version completely broke one of my cross-validation scripts and I suspect this was the issue. My personal opinion on this is that scale_C should default to False. As a user rather than developer of the library, if I tell the model to use some value of C, I expect it to use that value of C. I don't expect to have to read through all of the optional arguments to make sure I am not overlooking an "actually do what I told you to do" flag somewhere. On Mon, Apr 30, 2012 at 9:16 AM, Paolo Losi <[email protected]> wrote: > Hi Gael, > > On Sun, Apr 29, 2012 at 10:28 PM, Gael Varoquaux > <[email protected]> wrote: > >> >> It turns out that, for l2 penalizations, theory tells us that for >> prediction consistency (i.e. that under given hypothesis, the estimator >> learned predicts as well as an model knowing the true distribution) the >> penalty parameter should be kept constant as the number of samples >> grow. > > > Have you got any pointer to this theory? > > The tricky part in your reasoning IMHO is that I suspect that keeping > penalty > parameter costant is only justified when assuming gaussian priors on weights > really holds. That is something that cannot be generalized. > So, while it can be a reasonable default, considering C scaling a "bug" > doesn't really convince me. > > Ciao! > > Paolo > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > Scikit-learn-general mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/scikit-learn-general > ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
