Congratulations and some thinking about nectar.

[Xiaobo.Gu]

Hi,
Wonderful work you have done! After reading the white paper and
examples, I think the idea behind Nectar is similar as Cascading and
R. As you know, Cascading is providing a framework to data processing
without write raw MapReduce codes, and R is a framework to in-memory
data analytics. As a framework, they will provide as many primitive
data types and operations as possible. Algorithm developers will focus
on algorithm, but not raw programming. Another thing which I think is
the most important factor that R is so popular is that it provides a
package mechanism which allow algorithm developers to share their work
very efficiently.

What do you think then?

[madhukara phatak]

Hi

 I am big fan of R and they give fabulous library on statistics. Yes we know we lack lot of functions and we made it clear in FAQ too . But this is a humble start . We are planning to add more and more functions and we need lot of help from open source community too. But Nectar is not only about more and more functions.Its all about building a full fledged modeling framework  which covers a lot of areas. And we want to bring richness unix shell to our command utility too. So we know we have lot of work. So thank you for all your suggestion. In case of cascading it seems to general data processing API but in Nectar we are highly focused on modeling and machine learning.

--
Regards

Madhukara Phatak

[Xiaobo.Gu]

Will you provide up layer algorithms such as LogisticRegression, I
read somewhere you already have Neural Networks.

On Jul 24, 6:31 pm, madhukara phatak

<madhukarapha...@zinniasystems.com> wrote:
> Hi
>  I am big fan of R and they give fabulous library on statistics. Yes we know
> we lack lot of functions and we made it clear in FAQ too . But this is a
> humble start . We are planning to add more and more functions and we need
> lot of help from open source community too. But Nectar is not only about
> more and more functions.Its all about building a full fledged modeling
> framework  which covers a lot of areas. And we want to bring richness unix
> shell to our command utility too. So we know we have lot of work. So thank
> you for all your suggestion. In case of cascading it seems to general data
> processing API but in Nectar we are highly focused on modeling and machine
> learning.
>

[madhukara phatak]

This release provides the Linear multiple regression . But future releases will be providing the logistic regression and neural network work is already on the way. So soon you going to see in the framework.

--
Regards

Madhukara Phatak

[Xiaobo.Gu]

Great!

On Jul 24, 6:40 pm, madhukara phatak

[madhukara phatak]

Please try the framework and tell what you think . All your suggestions will be helpful to improve our framework.

--
Regards

Madhukara Phatak