Analytics Automated, A_A, is a Python/ Django application which allows you to turn computer software into RESTful webservices simply by configuring a web application. It is designed principally to allow users to make predictions with statistical models but it can be configured to run any software at all.
Operationalising statistical models and Data Science research is
hard, this framework simplifies this process. In fact you could turn
any piece of software in to a remote service users could call, even
rm -rf * (we don't advise you do this)
Technology lock-in is common in organisations. A_A lets you build models and data analysis pipelines for across different data analysis technologies. This leaves your organisation free to integrate or drop technologies and to hire the best talent without having to bet the farm on whichever of today's new hot Data Science technologies will still be around in a year's time.
A_A should be easy to understand for anyone with a reasonable knowledge of Python and a passing knowledge of Django. Set up is somewhat involved but once this is done configuring new services should be straightforward.
You should know Python 3. You should also be familiar with Django. A passing knowledge of Celery and RabbitMQ would not hurt either. We're not going to pretend this is a solution to instant and automatic Data Science success.
So far anything that can run on a standard unix commandline can be turned in to a service. As new backends are added to the commandRunner python package we'll add those to A_A. In the next month or so we'll be adding Grid Engine and R (via RServe)
If it runs on the commandline you can use it right now! If it runs in a specific modelling package we have plans to add support for many of them. Shortly, We plan to add Hadoop, Octave, Matlab and Mathematica support. We'd love to also add support for SAS Enterprise Server. Anything else is not currently on the roadmap. If you'd like to add support for anything specific please consider forking the commandRunner code base and adding the appropriate class.
Python has emerged as the de-facto scripting language for Data Science and research coding. We'd like this framework to be easily understood and adopted by these communities.
That's not really a question but yeah, this is the first production Python application we've built. We'll clean it up as we go but we'd love the code to be better please do fork the github repository and help us out. The tests could really do with some TLC.
A_A is distributed under the GPL v3.0
Please feel free to contact us via the project's github repo.
https://github.com/AnalyticsAutomated/analytics_automated
Yes you do. For now please use the webaddress for the project as per:
D.Buchan, 2015. https://analyticsautomated.github.io/