All Times EDT
In this talk, I will discuss the PCS framework from (Yu and Kumbier, 2020) (work-flow and documentation) that is built on the three principles of data science: predictability, computability and stability (PCS) and also discuss stability as a minimum requirement for interpretability. I will also cover interpretable machine learning through the PDR desiderata (Predictive accuracy, Descriptive accuracy and Relevancy) (Murdoch et al, 2019). I will illustrate the PCS framework for causal inference on heterogeneous treatment effects in randomized clinical trial data. In particular, we consider the problem of post-hoc discovery of subgroups with large subgroup average-treatment effects. (Based on joint work with Briton Park, Yan Shuo Tan, Mian Wei, Bin Yu and David Madigan.)