Greg, This is great! Sorry for making you work. The new help page is very informative. I am sure this will prove useful not just to me.
Questions: 1) How can I use real-valued descriptors, such as MOE-like descriptors for such modeling? Do I need to pick descriptors one-by-one or is there something like "AllDescriptors" which computes all of them in one pass? 2) Can you briefly say what each of the parameters to Grow() is? Some are mentioned in the text but I would like to make sure I know what everything is. Best regards, Igor On Tue, 2011-05-03 at 06:48 +0200, Greg Landrum wrote: > Hi Igor, > > On Mon, May 2, 2011 at 9:52 PM, Igor Filippov <igor.v.filip...@gmail.com> > wrote: > > > > Yes, actually for this project I'm interested in Python specifically! > > Time to learn me some new tricks :) > > Sad... I was kind of hoping to get the response: "oh? only in python? > never mind!" because then I wouldn't have to write the docs. ;-) > > > I was looking through the docs online but I cannot figure it out :( > > Yeah... that's the problem. > > The existing ML code is actually pretty easy to use (at least it used > to be... I haven't used it this way in a while) if you have a database > structured exactly the way it expects. When this is true, there are a > couple of command line tools that automate everything. However, I > don't expect people are actually going to be interested in doing this. > So there ought to be at least some docs. > > I will start a series of howto examples on the wiki and label them with "ML": > http://code.google.com/p/rdkit/w/list?q=label:ML > the first is here: > http://code.google.com/p/rdkit/wiki/BuildingModelsUsingFingerprints1 > > I'm going to focus on tree predictors (i.e. bags of decision trees and > random forests), because that's the area where the RDKit is the > strongest. > > If you're interested in more flexibility in terms of type of model, > it's probably best to use the RDKit in combination with R (via rpy2, I > haven't done this), or knime. > > -greg ------------------------------------------------------------------------------ WhatsUp Gold - Download Free Network Management Software The most intuitive, comprehensive, and cost-effective network management toolset available today. Delivers lowest initial acquisition cost and overall TCO of any competing solution. http://p.sf.net/sfu/whatsupgold-sd _______________________________________________ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
