Data scientists are often asked causal questions, like "what would happen if we did X?", but are often can't run experiments to answer them. There's a mature literature on methods for causal inference, but it takes a big time investment to develop the skills to understand the methods and pitfalls. Ill argue for the need to invest this time, and give a quick overview of methods and tools, including my causality python package.
Adam has a PhD in physics and has spent the past several years doing data science at BuzzFeed. He's now heading Barclays's research data science team, and teaching applied causal inference in Columbia's data science institute.