Great article! You can't get insights without elegant data pipelines though. And modern data science graduates don't even know SQL and work with Pandas dataframes only :) If something goes wrong with prediction Data Scientist would def start blaming data crying it's not clean enough.