That's really encouraging to see my research ideas getting into sklearn! Thanks so much for taking the time to do this :)
I'm definitely interested to hear how it works for people in various applications. I think that generally for 3 or more dimensions, random search should match or exceed grid search. I'd like to also add that for applications with maybe <= 5 dimensions, it might be best to use some of the quasi-random sequences like Sobol or Niederreiter or something. I'm not sure what is the best way to access these things from Python though. The GSL has them, but the license is too strict right? I had hoped to find them in scipy.integrate, but I don't see any of those kinds of sequences there. I see that in [1] there are some interesting LGPL codes, can they be brought into scipy? [1] http://people.sc.fsu.edu/~jburkardt/py_src/sobol/sobol.html On Sat, Oct 6, 2012 at 7:37 AM, Andreas Mueller <[email protected]> wrote: > Hi everybody. > Last weekend I created a draft of a random hyper parameter search a la > James. > It doesn't do tree-structured parameter spaces and doesn't do the Parzen > estimate > (yet), just random sampling. > Feedback would be very welcome, as I want to use this as a basis for a > better grid search > visualization PR. > You can find my PR here: > https://github.com/scikit-learn/scikit-learn/pull/1194 > > Cheers, > Andy > > ------------------------------------------------------------------------------ > Don't let slow site performance ruin your business. Deploy New Relic APM > Deploy New Relic app performance management and know exactly > what is happening inside your Ruby, Python, PHP, Java, and .NET app > Try New Relic at no cost today and get our sweet Data Nerd shirt too! > http://p.sf.net/sfu/newrelic-dev2dev > _______________________________________________ > Scikit-learn-general mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/scikit-learn-general ------------------------------------------------------------------------------ Don't let slow site performance ruin your business. Deploy New Relic APM Deploy New Relic app performance management and know exactly what is happening inside your Ruby, Python, PHP, Java, and .NET app Try New Relic at no cost today and get our sweet Data Nerd shirt too! http://p.sf.net/sfu/newrelic-dev2dev _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
