The examples can be found in the documentation. They are, admittedly, a bit difficult to find, and we're working on that.
The developer guide can be found here.
For example to see an adaboost batch example, please navigate on the left navigation pan:as follows:
->Python* Language Application Programming Interface Reference for Intel® Data Analytics Acceleration Library
On the right under "References" click on "AdaBoost algorithm description and usage models". From there navigate to "Batch Processing" where you will find a link to the python example at the bottom of the page.
It may be easier to go here and download the reference guide. Then, in the daal_python_api folder you can find all the examples with the "-example.htm" suffix.
I hope that helps,
We understand that Python code examples availability is now the gap that needs to be addressed. Certainly availability of example code through the documentation is not the primary channel where examples should come from.
I need your help to understand the best way to deliver examples to you. Should it be part of pydaal package? Should it be separate package with examples? Should it be available on github? Should it come with Intel Distribution package in one of python directories? All above?
Give me some hint what options would work best for you.
I have a few Jupyter notebooks about using pyDAAL: https://github.com/zhangzhang10/pydaal-tutorials
There are just a few at this moment but it will grow over the time. And, I believe these are easier to consume by new users.
Hi Zhang. Thanks! That's actually what I had in mind of doing myself.
I was going to suggest to Sergey to provide jupyter notebooks for the examples, with the additional idea of Intel hosting the notebooks on a server or cluster using jupyterhub. Users could log in with their serial number and could play around with the examples without worrying about setting up their own systems. It could also be a way of people eventually trying out the Xeon Phi and maybe even FPGAs.
Sort of along the lines of this, https://zonca.github.io/2016/04/jupyterhub-image-sdsc-cloud.html
I'm looking into this myself for a classroom setting where I'd do a presentation to a class on gpu computing using a notebook on a remote server with gpu instance, then give them assignments where they modify the notebook. pydaal will be on that system as well.
Looks like my replies are still being moderated... I guess you'll see this soon enough.
Hmm... looks like my earlier reply to Christopher and Sergey went missing. I'll see if I can remember what I said.
Christopher. I did find the python examples in the documents, but for example datastructures_aos.py can't find the utils folder on import so it looks like the python examples can't run without a download base. I had attempted to download pydaal with a community license but never got a link in the email I was sent. I eventually tried registering for a license id again and this time got a download link.
Sergey, thanks for asking! I think it may be better to keep pydaal on the lean side with an option to download the examples from github where the other examples are. I think the Intel Distribution package should have the python examples. My nickle anyway...