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: Using Samoa/Samza to perform root cause analysis
Date Wed, 04 Feb 2015 11:26:26 GMT
Hi Shekar,

As I said, I am not familiar with the problem so I am not able to help you
on how to frame it.
However, given that you are just starting out I would first try with some
single-machine, batch machine learning library such as WEKA or scikit-learn.
Once you have developed an understanding of how to frame the problem and
which features you want to use, then you can start thinking about problems
of scaling and streaming.

Cheers,


--
Gianmarco

On 2 February 2015 at 16:54, Shekar Tippur <ctippur@gmail.com> wrote:

> Gianmarco
>
> I am trying to get over the initial teething issues.
> Can you please give me initial pointers on how to frame it as a
> classification problem?
> Also, how do I feed training data to the model?
> Appreciate your help.
>
> Shekar
>  On Feb 2, 2015 3:31 AM, "Gianmarco De Francisci Morales" <gdfm@apache.org
> >
> wrote:
>
> > Hi Shekar,
> >
> > I am not sure whether it is a typical use case.
> > If you can frame it into a classification problem, then yes SAMOA can
> help
> > you.
> >
> > I am not aware that it has been done, but I am not familiar with the
> area.
> >
> > If you want to apply supervised ML, you need to decide which features you
> > want to extract and how to get training data.
> >
> > Cheers,
> >
> >
> > --
> > Gianmarco
> >
> > On 27 January 2015 at 23:44, Shekar Tippur <ctippur@gmail.com> wrote:
> >
> > > Gianmarco,
> > >
> > > Thanks for pointing me to the right email group to address any
> questions
> > on
> > > Samoa.
> > >
> > > My name is Shekar.
> > > I am working on a solution to enrich events (mainly alerts) going into
> > > operation centre. We have data on
> > > 1. Prior postmortem results
> > > 2. Data on best practices.
> > > 3. Data on typical root cause on the web (like null pointer exception,
> > Out
> > > of memory.. etc)
> > >
> > > I want to enrich each event (alert) with possible root cause from these
> > > sources.
> > >
> > > Had couple of questions:
> > > 1. Is this a typical use case for Samoa?
> > > 2. Any idea if this has been done before?
> > >
> > > I want to apply ML to the incoming events and learn from it.
> > >
> > > Appreciate any pointers.
> > >
> > > - Shekar
> > >
> >
>

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