Re: Implementing Naive Bayes on Top of Storm

[Gianmarco De Francisci Morales <gd...@apache.org>]

Hi,

Yes, SAMOA has a NB implementation. However, the implementation works only
for ensemble methods right now (
http://samoa.incubator.apache.org/documentation/Bagging-and-Boosting.html).
You can have a look at the code here:
samoa-api/src/main/java/com/yahoo/labs/samoa/learners/classifiers/NaiveBayes.java
<https://github.com/apache/incubator-samoa/blob/94f6f7652ad19c0160622a5d34cfa80119cca07a/samoa-api/src/main/java/com/yahoo/labs/samoa/learners/classifiers/NaiveBayes.java>

If you want a single NB method, it can also be done but requires a bit of
coding to integrate the existing NB into a SAMOA learner, and handle edge
cases. We have plans to do it anyway so if you want to tackle the task it
would be a welcome addition.

What you need to write is a way to convert your raw data (text) in SAMOA
instances.
We are working on a preprocessing pipeline but it's still not ready.

Do not hesitate to ask if you need more guidance.

Cheers,

--
Gianmarco

On 31 March 2015 at 17:51, Raman Jhajj <bo...@gmail.com> wrote:

> Hi Everyone,
>
> I am new to SAMOA and starting to explore it. I have a requirement to
> implement multilabel classification preferably using Naive Bayes
> Multinomial. This need to be implemented on top of Storm and classification
> need to be done on the incoming stream of text data.
>
> I am clueless about how to proceed on this. Can anyone help me out with
> some guidence or pointers if this is possible using SAMOA?
>
> Do I need to write Naive Bayes Multinomial algo myself or is it already
> implemented in SAMOA? If I need to write, how can I do that.
>
> Please help me in this. Thanks a lot
>
> --
> Kind Regards,
>
> *Ramaninder Singh*
>