Learning Data Mining with Python - Second Edition

Automated Learning

URL: http://rhiever.github.io/tpot/

URL: https://github.com/automl/auto-sklearn

It's almost cheating, but these packages will investigate a wide range of possible models for your data mining experiments for you. This removes the need to create a workflow testing a large number of parameters for a larger number of classifier types, and lets you focus on other things, such as feature extract--still critically important and not yet automated!

The general idea is that you extract your features and pass the resulting matrix onto one of these automated classification algorithms (or regression algorithms). It does the search for you and even exports the best model for you. In the case of TPOT, it even gives you Python code to ...