mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Koobas <koo...@gmail.com>
Subject Re: Using Mahout for low-volume data
Date Mon, 15 Jul 2013 15:37:25 GMT
Is a factorizing recommender a better idea for low volume data in general?


On Mon, Jul 15, 2013 at 11:35 AM, Ted Dunning <ted.dunning@gmail.com> wrote:

> With such small data, this sounds (without thinking too much) like you are
> doing reasonably well with LLR similarity.
>
> Have you tried a factorizing recommender?
>
>
> On Sun, Jul 14, 2013 at 10:49 PM, Jayesh <jayesh.sidhwani@gmail.com>
> wrote:
>
> > Hi Ted,
> >
> > Thanks for the reply.
> > My training data could have around 100k users and around 1k items. The
> data
> > is sparse (I have a boolean affinity - the user either bought the item or
> > did not)
> >
> > PS: I have been playing around with a sample code, using Loglikelihood
> > Similarity to get a 24% precision, is this a par score?
> >
> >
> >
> > On Mon, Jul 15, 2013 at 10:58 AM, Ted Dunning <ted.dunning@gmail.com>
> > wrote:
> >
> > > Mahout will work fine for smaller data sizes.
> > >
> > > Collaborative filtering can be difficult in general with small data,
> > > however.
> > >
> > > How many users and how many items?  How many actions?
> > >
> > >
> > > On Sun, Jul 14, 2013 at 10:22 PM, Jayesh <jayesh.sidhwani@gmail.com>
> > > wrote:
> > >
> > > > Hello,
> > > >
> > > > I am exploring the collaborative filtering algorithms in Mahout to
> > build
> > > a
> > > > recommendation engine.
> > > >
> > > > I had recently gone for a Big Data conference where the speakers
> > > suggested
> > > > that using Mahout is overkill for anything that doesn't have some
> > > terabytes
> > > > of training data.
> > > >
> > > > I tried to google some cases on that, but no help, so turned to this
> > > > thread.
> > > >
> > > > Would you suggest me using Mahout in my case when the training data
> > might
> > > > not be in terabytes, but some gigabytes, perhaps few 100s of
> megabytes?
> > > > If not, do you suggest any other approach?
> > > >
> > > > Thank you.
> > > >
> > > > --
> > > > Best Regards,
> > > >
> > > > Jayesh
> > > >
> > >
> >
> >
> >
> > --
> > Best Regards,
> >
> > Jayesh
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message