How do we do Agile in Data Science?

Here is my notes from the excellent article on Agile process for Data Science - Agile in Data Science

How is data science different from software engineering?

Data Science workflow is a mix of waterfall and iterative approaches. Typically it involves:

  • Data exploration
  • Algorithm research
  • Algorithm trails
  • Model improvements
  • Data exploration

How to succeed with Agile?

The best way to bring predictability to your Data Science team is to interlace the research and operational work. So in every sprint, you’ll do both, but on different topics. You’ll research ahead and do the operation for the area/opportunities you already got successful outputs from your research.

Typically it looks like this for a given epic