samoa-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gianmarco De Francisci Morales <g...@apache.org>
Subject Re: Distributed stream clustering Limitations
Date Mon, 13 Feb 2017 08:52:11 GMT
Hi,
Please note that the clustering algorithm has not received as much
attention as the classification/regression ones.
1) The precise numbers can be different, but it needs some initial number
of events.
2) It can be changed.
3) It can be changed.

It should be fairly straightforward to have parameters for these quantities.
If you want to contribute a patch, we would be happy to review and merge it.

Cheers,

-- Gianmarco

On Sun, Feb 12, 2017 at 6:27 PM, Nicolas Kourtellis <nkourtellis@gmail.com>
wrote:

> Hi Jayan,
>
> Thanks for testing the stream clustering methods.
> I assume you should be able to modify those numeric variables you point
> out,
> try out different numbers based on the datasets you have in mind testing
> and of course recompile your local code with maiven and try it out.
>
> Maybe some of these parameters should be more appropriate to be passed down
> and be defined in the command line execution?
>
> Anyhow, after your testing, let us know if you have some suggestions.
>
> Thanks,
>
> Nicolas
>
>
>
> On Tue, Feb 7, 2017 at 7:13 AM, Jayan Vidanapathirana <
> jayancv.13@cse.mrt.ac.lk> wrote:
>
> > Hi all,
> >
> > In SAMOA distributed stream clustering I found some limitations
> >
> > 1) Do it needs 12000 events to initiate the model?
> > 2) Sample frequency fixed to 1000
> > 3) It outputs cluster center only after processing 10000 events
> >
> > Is there any way to fix these issues?
> >
> > --
> > Best Regards
> >
> > Jayan Vidanapathirana
> > Undergraduate (CSE '13 Batch )
> > Computer Science & Engineering Department
> > University of Moratuwa.
> > jcvidanapathirana@gmail.com
> > [image: https://lk.linkedin.com/in/jayancv]
> > <https://lk.linkedin.com/in/jayancv>
> >
>
>
>
> --
> Nicolas Kourtellis
>

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