Great article! I believe that every data science project needs to start with a deep understanding of the business problem and its correct translation into analytical problem.

A lot can go wrong when this translation is done by a person who is not both tech/data and business savvy.

I’ve recently summarised my opinions here.