At Users Conference San Francisco, we offered a lab that showed how you can connect PI to Azure Machine Learning (Azure ML) platform and get more value out of PI Data. This video series by O'Reilly teaches you more on Azure ML side, how it works, how you can leverage existing modules to do machine learning, build your own algorithms using R and/or Python, and publish your creations as a web service for others to use.
I'll need to check out the O'Reilly videos. I've been disappointed by the lack of learning resources for machine learning. What I do find seems to assume basic machine learning experience. It's difficult to find something to at least get you to novice. I have found this Manning book to be very helpful, although it's a bit agnostic in that it doesn't provide information about machine learning platforms (Azure, Tensor Flow, etc.).