Examples

The examples show many different ways WhyNot can be used to study questions in causal inference and sequential decision making. These experiments are meant to be illustrative rather than exhaustive. We encourage you to add more examples to the library!

Causal Inference

• Heterogeneous treatment effects
• Interaction between experimental units
• Mediation analysis
• Observed and unobserved confounding
• Sensitivity analysis
• Instrumental variable estimation
• Regression discontinuity design
• Graphical methods with automatically generated causal graphs
• Causal graph discovery

Decision Making in Dynamics

• Delayed impact of fair machine learning
• Performative prediction

Reinforcement Learning

• Design optimal HIV treatment policies
• Prevent population collapse in World3
• Design policies to control the spread of Zika virus