Most media coverage of innovation in data science focuses on machine learning, but in this post, Thomas Vladeck, a Managing Director at Gradient Metrics (a quantitative market research firm), describes how advances in Bayesian inference and causal inference are changing how data practitioners work. For example, he describes how Hamiltonian Monte Carlo and Automatic Differentiation Variational Inference are enabling data scientists to write custom models to more accurately estimate uncertainty. He also discusses new approaches to causal modeling of observational data, like DAGitty – which allows users to easily draw and analyze causal diagrams.