Data-Driven Decision Making with Entropic Risk Measure

Explore a seminar on data-driven decision making using entropic risk measures. Delve into the challenges of decision-making under uncertainty, where decision-makers may select options with higher expected losses. Examine the entropic risk measure as a widely-used tool for accounting for risk aversion in decision-making processes. Investigate the limitations of using finite samples to estimate true entropic risk and learn about two innovative procedures based on optimal transport and extreme value theory. Discover how these methods use Gaussian mixture models and bootstrapping to correct bias in entropic risk estimation. Study the application of distributionally robust optimization with entropic risk measures, including the use of debiasing in cross-validation for tuning ambiguity set parameters. Gain insights into practical applications of these techniques in project selection, portfolio optimization, and insurance pricing problems, demonstrating how debiasing approaches can lead to lower risk outcomes in real-world scenarios.