Causal Estimates using Machine Learning
Hemant Ishwaran · 2025
This chapter delves into the concept of causation and discusses classical estimators used for the estimation of causal effects from observational data. We also explore advanced machine learning methods for estimating individual treatment effects (ITE), an extension of the population average treatment effect (ATE) used in classical analyses, using the concept of virtual twins (digital twins). Through this exploration, the chapter aims to provide a comprehensive understanding of how these methodologies contribute to the accurate assessment of causal relationships and the assumptions necessary to achieve accurate estimates using observational data.