题目：Data Driven Forecasting and Optimization for Random Yield Problems: Theory and Applications to Agro Business
主讲人：Mahesh Nagarajan，Chair Professor, the Sauder School of Business, the University of British Columbia, Canada
We revisit the well known random yield stochastic inventory control model. We propose a modification of the linear inflation rule heuristic and use it to derive content approximation guarantees for a class of random yield network problems. We also study this problem when the demand and yield forecasts are not well understood, using a combination of min max and sampling average approximations. We use these results to (1) provide a easy to use tool that performs well for large instances of this problem that appear in industry in the aggro business; (2) to derive bounds on the "amount of forecasting" that needs to be done to get provably close to the optimal solution and (3) understand the impact of optimization and forecasting in a product network.
Mahesh Nagarajan is the Alumni Chair Professor of Stochastic optimization at the Sauder school of business at the university of british columbia. He is also the division chair for the Operations research and logistics group at UBC. His research interest include applications of optimization to operations management problems as well as cooperative game theory. He serves as an associate editor for MS, OR, POMS and department editor for ORL and has served as a associate editor for Math of OR. He has won several research awards and publishes regularly in the top journals in our field.