SIG Causal Data Science
Much of scientific research aims to gain insight into causal relations: Does exposure to a particular factor cause disease onset? Was the introduction of a government policy successful in achieving a particular aim or not? How should we intervene in a system to achieve some outcome, and what effect can we expect that intervention to have? While randomized control trials are understood to be the gold standard for answering causal research questions, they are often impractical, inefficient, unethical or impossible to conduct in different scientific contexts. Data science methods have been a game changer in performing prediction and classification tasks across a number of scientific domains, but it is often unclear whether, when and how these methods can be leveraged to help answer causal research questions.
This special interest group will bring together researchers who develop and apply data science methods designed to yield causal information from different data sources. A variety of causal modeling methods have been developed in recent years, notably in fields as diverse as computer science, epidemiology and economics. Our aim is to bring together both substantive and methodological researchers from a wide variety of different relevant disciplines who have experience with (developing) these approaches. In so doing we hope to foster new collaborations, promote novel applications of causal data science techniques, and explore new methodological developments.
The SIG causal data science will organize a regular causal data science reading group, open to all researchers at the UU and UMC Utrecht. This group will focus on discussing both classic texts and new developments from the field of causal modeling, as well as inviting researchers in the field to present and discuss their own causal data science research.