Intel® oneAPI Data Analytics Library
Learn from community members on how to build compute-intensive applications that run efficiently on Intel® architecture.
Announcements
The Intel sign-in experience has changed to support enhanced security controls. If you sign in, click here for more information.
216 Discussions

What programming languages does this library support?

Ying_H_Intel
Employee
480 Views

Question : 

What programming languages does this library support?

0 Kudos
10 Replies
Zhang_Z_Intel
Employee
480 Views

Intel DAAL currently provides C++ API and Java API. Depending on customer's demand we may add support for other programming languages in future. Let us know your preferences!

gaston-hillar
Black Belt
480 Views

@Zhang Z Python is extremely popular as a programming language for data science. Are there any plans to include Python wrappers in the final version?

Sergey_M_Intel2
Employee
480 Views

Dear Gaston,

We're considering Python API. It's good time to share your requirements, if any.

* What Machine Learning, Scientific, and Data Management Python packages do you currently use?

* Are there any compatibility requirements with other packages? If yes how much performance can you tolerate?

* Other requirements?

Thank you,

Sergey

gaston-hillar
Black Belt
475 Views

@Sergey,

That's great news. I'm sure there is a very important number of Python developers that will love to have a Python API for this library.

Let me provide you the information you requested. The most popular packages I've been working with in many projects are:

Machine Learning: Scikit-Learn

Scientific Package: Sci-Py

I do believe compatibility with these two packages would be extremely useful.

Let me know any additional information that might be helpful to provide.

gaston-hillar
Black Belt
475 Views

@Sergey,

I also worked with Pydoop and Pig. However, not sure how popular they are amongst Python developers that work with Python and Hadoop. I think Jython is one of the most populars when working with Hadoop.

 

Sergey_M_Intel2
Employee
475 Views

Hi Gaston,

This is useful. Scikit-Learn, NumPy/SciPy, Pandas are on our radar. In terms of compatibility, what kind of compatibility are you looking for? For instance, full API compatibility is impossible due to DAAL supports greater variety of usage scenarios than Scikit-Learn. Second aspect is that compatibility always means tradeoff and the key question is how much do you want to tolerate, e.g. in terms of performance or may be in other categories.

That will help us in making right tradeoffs

Regards,

Sergey

gaston-hillar
Black Belt
480 Views

@Sergey,

 

You are right. I've been diving deep on DAAL in the last days as I discovered it a week ago and DAAL supports greater variety of usage scenarios compared with Scikiy-Learn.

IMHO, the most important thing is to have a Python wrapper.

When developers choose Python, they accept a performance tradeoff. Python is cool, but not fast as C / C++. Thus, IMHO, it makes sense to have a performance tradeoff.

 

Sergey_M_Intel2
Employee
480 Views

Got it! Thank you Gaston!

gaston-hillar
Black Belt
480 Views

@Sergey,

In case you need additional details that might help, you can drop me an e-mail to gastonhillar at the domain hotmail dot com.

I'll be working testing DAAL in different scenarios and with its supported programming languages as it is a library in which I'm very interested in.

Sergey_M_Intel2
Employee
480 Views

Hi Gaston,

Thank you! Much appreciated! We'll discuss with engineering teams and come back with questions.

Thank you,

Sergey 

Reply