I've been following some machine learning tutorials for Python by Sentdex on YouTube. He's notable for his prolific output of Python tutorials on various topics and for making the topics reachable to all audiences.

He has tutorial series in financial analysis, python skills and every sort of interesting thing Python can do.

Here's his machine learning series that I'm working through:

Quandl

Sentdex pulled sample data from Quandl using their Py module. Their website includes an API section to get you started pulling the data using whatever language or software you prefer. What is Quandl? Simply put, it's one of the coolest data providers I've come across! They mainly deal in financial data, including End Of Day stock results and fundamental indicators of company and economic health. That's just the beginning, they made a name for themselves by attracting alternative data as different as satellite measurements of global oil tank storage levels to national UFO sightings (yes the last one is real, but the jury is out on whether it's real.)

The data I was most interested in was housing data from Zillow, Federal Reserve data on the economy, European Central Bank records and even China Macroeconomic and Industrial Data. It's invigorating to see 'live' data that I can easily pull in and play with and not just read summary articles by others. Will I gain insights? At this point I highly doubt it, however it's a great playground for machine learning and maybe one day I'll contribute some ideas to how things work.

Heroku

What is Heroku? It's technically referred to as a PaaS (Platform as a Service), meaning they provide not just a server but a managed software environment to host your apps. They take care of scaling (based on your specifications), database management and all the background infrastructure problems which could take you down. All you do is bring your app and manage it.

I'm looking at some new web service projects and Heroku stood out for its popular usage. If languages can be lower or higher level based on their complexity then PaaS could be called higher level for doing more to manage infrastructure than IaaS providers do. Infrastructure as a Service (IaaS) provides the server backbone to host your own software. PaaS providers handle software licenses and other sysadmin duties for you. Well maybe you already know this.

I followed Heroku's free tutorial which featured downloading their CLI and uploading a premade Django web app. It honestly wasn't too complicated in concept and pretty familiar to my blog development process. Developing an app, applying git version control to track progress and push updates. The interesting parts were Heroku's management software, hosting your app locally, spinning up your one Free dyno to serve it from Heroku. All in all they seem to do a good job of separating the infrastructure from app development, and I wouldn't mind to work with them.

That was the weekend!