machine learning

For me, the highlight of Tuesday was the keynote by Susan Athey “Learning Personalized Policies”. Susan is an incredibly versatile economist who is probably best known for merging two very different streams of research: causal inference in economics and machine learning in computer science. If you want a simple intro to this topic, this Science article is a good start. To put it simply, causal inference tries to prove that “A causes B” using any means possible: ideally through a controlled experiment, but if not possible, then through other tricks (instrumental variables, matching etc.). Machine learning approaches often use the same exact data or application to answer a different question: “how can I best predict what happens”. For a long time, these two communities existed separately but they are increasingly overlapping and both approaches are needed for real-work decision making. If you are not familiar with Susan’s work, I highly recommend this source where you will find her teaching materials, videos of her tutorials and all relevant papers with code. This is really the future of data analysis.