At what point in your data pipeline do you check that your data adhears to constraints on it? Is it embedded in the code that performs the ingestion, at the point of the analysis or, prehaps, do you have a utility that allows you to view data in your database that do not conform to constraints on it?
I have worked in several environments as a statistician/data scientist where the project management for my projects ranged from 'let me know when your done' to burndown charts and Agile. This got me thinking: what project management structure do other data scientist/statisticians use and is there a best set of practicies? For me, at the very least, I would like to see source control management, data dictionaries and explicit documentation of the hypotheses of the project.
