Yes indeed, as I said at the time, much of the forest code could be reused to implement a pure averaging meta-estimator.
The main thing that makes BaseForest tree-specific is that it precomputes X_argsorted such that it is computed only once for all trees and inject it into the fit method of the base trees. But besides that, any base estimator could actually be used. Gilles On 20 January 2012 21:17, Lars Buitinck <[email protected]> wrote: > 2012/1/20 Andreas <[email protected]>: >> In how far is #491 tree specific? >> This is parallelization over different boot strap samples. >> Or am I missing something there? >> Feature importance (#478) is not as generic but >> "just" relies on feature importance from the base classifier, >> right? >> Or did I miss something there, too? > > I didn't follow that PR and was surprised that it depended on decision > trees when I saw it after the merge. Prinzie and Van den Poel [1][2] > build random "forests" of multinomial LR classifiers, calling the > result random multinomial logit. I've yet to read it thorougly, so I'm > not sure they use the exact same (meta-)algorithm that our forests do. > > [1] http://www.sciencedirect.com/science/article/pii/S0957417407000498 > [2] https://en.wikipedia.org/wiki/Random_multinomial_logit > > -- > Lars Buitinck > Scientific programmer, ILPS > University of Amsterdam > > ------------------------------------------------------------------------------ > Keep Your Developer Skills Current with LearnDevNow! > The most comprehensive online learning library for Microsoft developers > is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3, > Metro Style Apps, more. Free future releases when you subscribe now! > http://p.sf.net/sfu/learndevnow-d2d > _______________________________________________ > Scikit-learn-general mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/scikit-learn-general ------------------------------------------------------------------------------ Keep Your Developer Skills Current with LearnDevNow! The most comprehensive online learning library for Microsoft developers is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3, Metro Style Apps, more. Free future releases when you subscribe now! http://p.sf.net/sfu/learndevnow-d2d _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
