Another issue that comes from times to times is the fitness of the sklearn
API wrt to recommendation tasks.
I believe it's pretty good if one has to manipulate - e.g.factorize -
(item, user) matrices, but it falls short when dealing with explore/exploit
scenarios.
An example of that is the bandit [1] family of algorithms, where one knows
the payoff of an action iff the action is chosen by the algorithm as the
next step.
[1] http://en.wikipedia.org/wiki/Multi-armed_bandit
2013/10/9 SUJIT PAL <[email protected]>
> I believe there is already a recommender framework in the scikits family
> already called crab?
> http://muricoca.github.io/crab/
>
> Few days back, one of the committers to sklearn spoke about the fact that
> he detected code in crab that looked like his own. Given that there is so
> much reuse, would it make sense for crab (aka scikits.recommender) to use
> sklearn as a dependency and build any recommender system specific code in
> crab?
>
> -sujit
>
> On Oct 8, 2013, at 2:51 PM, Joel Nothman <[email protected]> wrote:
>
> > On Tue, Oct 8, 2013 at 11:32 PM, Olivier Grisel <
> [email protected]> wrote:
> > 2013/10/8 Gael Varoquaux <[email protected]>:
> > > On Tue, Oct 08, 2013 at 07:47:40AM +0200, Gilles Louppe wrote:
> > >> Unfortunately, algorithms for recommender systems are not planned in
> > >> scikit-learn in the short or mid-term.
> > >
> > > Indeed in the short term, but are we sure that we want to close the
> door
> > > to contributions implementing standard approaches for recommender
> > > systems?
> >
> > +1 for encouraging pull requests that implement recsys building blocks
> > (e.g. matrix factorization) that fit the scikit-learn API (fit and
> > partial_fit + predict or transform) and work with standard input
> > datastructures (e.g. input data is a scipy.sparse matrix or numpy
> > array).
> >
> > We don't want frameworkish code that hard code recsys specific
> > concepts (e.g. users and items) in the API though.
> >
> > I'm not familiar enough with recommender systems to understand whether
> any of the existing matrix factorisations apply. Is this more a matter of
> presenting an example of their application to this task?
> >
> > - Joel
> >
> ------------------------------------------------------------------------------
> > October Webinars: Code for Performance
> > Free Intel webinars can help you accelerate application performance.
> > Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most
> from
> > the latest Intel processors and coprocessors. See abstracts and register
> >
> >
> http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk_______________________________________________
> > Scikit-learn-general mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
>
>
>
> ------------------------------------------------------------------------------
> October Webinars: Code for Performance
> Free Intel webinars can help you accelerate application performance.
> Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most
> from
> the latest Intel processors and coprocessors. See abstracts and register >
> http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk
> _______________________________________________
> Scikit-learn-general mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
>
------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
