Introduces a data-driven approach to causal modeling for decision support that emphasizes empirically testable assumptions. The talk discusses how causal Bayesian networks and influence diagrams can be learned and validated from observational data, with applications to risk analysis and decision-making under uncertainty.
