Thanks Supun!

>From what I understood, way it works is.


   1. I take a dataset
   2. I do various statistical tests and decide AR, MA, ARIMA and estimate
   parameters.
   3. Then apply the built model to predict the next value


#1 and #2 we should do in batch fashion using Spark ( and add a Path in ML
Wizard for user to do it). However, #3 we need to do from CEP, as the data
arrives. Is that possible to take the code from the project to do #3 as
well?

--Srinath


On Thu, Aug 13, 2015 at 3:39 AM, Supun Sethunga <[email protected]> wrote:

> [Moving the discussion to arch@]
>
> Hi all,
>
> I looked into their code and infact they do have ARIMA implementation [1]
> , though its not there in the documentation. Not sure whether its still
> under development. Anyway, they also do have some statistical tests (Durbin
> watson test, Breusch-Godfrey Test) which we might need.
>
> So, regarding what we can do in ML.. think we can support both AR and MA
> separately (for anyone who is interested), and ARIMA as the generalized
> version. I created a very Basic summary on ARIMA model [2]. Will improve
> this, going forward :)
>
> [1]
> https://github.com/cloudera/spark-timeseries/blob/master/src/main/scala/com/cloudera/sparkts/ARIMA.scala
> [2]
> https://docs.google.com/a/wso2.com/document/d/1pcGT620Fjms9gxmyn85mNP93oye_5xxCBhtqvE6iAAw/edit?usp=sharing
>
> Thanks,
> Supun
>
> On Wed, Jul 29, 2015 at 2:21 AM, Supun Sethunga <[email protected]> wrote:
>
>> Sure Srinath. By the look of it, seems they do have "Exponentially
>> weighted moving average" and  "Auto-regressive models". Will do bit of
>> research on whether they support ARMA/ARIMA stuff (combination of the
>> earlier two) which are the most widely used models.
>>
>> Thanks Nirmal for the Link btw :)
>>
>> On Wed, Jul 29, 2015 at 11:30 AM, Srinath Perera <[email protected]>
>> wrote:
>>
>>> Supun, please checkout and send notes.
>>>
>>> --Srinath
>>>
>>> On Wed, Jul 29, 2015 at 11:29 AM, Seshika Fernando <[email protected]>
>>> wrote:
>>>
>>>> Ah, this is superb. It has the VaR stuff that I wanted to do also. :)
>>>>
>>>> On Wed, Jul 29, 2015 at 11:07 AM, Nirmal Fernando <[email protected]>
>>>> wrote:
>>>>
>>>>> A library for financial and time series calculations on Apache Spark
>>>>> https://github.com/cloudera/spark-timeseries
>>>>>
>>>>> --
>>>>>
>>>>> Thanks & regards,
>>>>> Nirmal
>>>>>
>>>>> Associate Technical Lead - Data Technologies Team, WSO2 Inc.
>>>>> Mobile: +94715779733
>>>>> Blog: http://nirmalfdo.blogspot.com/
>>>>>
>>>>>
>>>>>
>>>>
>>>
>>>
>>> --
>>> ============================
>>> Blog: http://srinathsview.blogspot.com twitter:@srinath_perera
>>> Site: http://people.apache.org/~hemapani/
>>> Photos: http://www.flickr.com/photos/hemapani/
>>> Phone: 0772360902
>>>
>>
>>
>>
>> --
>> *Supun Sethunga*
>> Software Engineer
>> WSO2, Inc.
>> http://wso2.com/
>> lean | enterprise | middleware
>> Mobile : +94 716546324
>>
>
>
>
> --
> *Supun Sethunga*
> Software Engineer
> WSO2, Inc.
> http://wso2.com/
> lean | enterprise | middleware
> Mobile : +94 716546324
>



-- 
============================
Blog: http://srinathsview.blogspot.com twitter:@srinath_perera
Site: http://people.apache.org/~hemapani/
Photos: http://www.flickr.com/photos/hemapani/
Phone: 0772360902
_______________________________________________
Architecture mailing list
[email protected]
https://mail.wso2.org/cgi-bin/mailman/listinfo/architecture

Reply via email to