The new science of cause and effect, with reflections on data science and artificial intelligence

摘要

The past three decades have seen the development of powerful tools for modeling and computing causal relationships which may have major impact on data science. My talk will illustrate how these tools work in seven tasks:1.Encoding causal assumptions in transparent and testable way2.Predicting the effects of actions and policies3.Computing counterfactuals and finding causes of effects4.Computing direct and indirect effects (Mediation)5.Integrating data from diverse sources.6.Recovering from missing data7.Discovering causal relations from dataA friendly, non-technical account of these ideas is available in: “The Book of Why: the new science of cause and effect,” Judea Pearl and Dana MacKenzie, (Basic Books, 2018). http://bayes.cs.ucla.edu/WHY/