Spring 2007 Seminar Series

DATE: Thursday, February 15, 2007
LOCATION: E40-298
TIME: 4:15pm
Reception immediately following in the Philip M. Morse Reading Room, E40-106

SPEAKER:
Assaf Zeevi

TITLE
Two Problems Involving Sequential Estimation and Optimization

ABSTRACT
This talk illustrates some ideas in sequential estimation and dynamic optimization under uncertainty; the following two problems provide motivation and backdrop for the topics that will be discussed.

Problem 1: A marketer has the ability to dynamically allocate one of several unique marketing messages to each individual consumer based on some knowledge of their profile and past interactions. The problem is to maximize the expected response rate over the course of a marketing campaign by suitably balancing exploration (learning consumer response) and exploitation (assigning the ``best'' message) objectives.

Problem 2: A revenue manager wishes to infer real time changes in mean market response, or willingness-to-pay distribution, over the course of a selling season. (This in turn supports judicious price changes, promotions discounts etc.) The problem is to sequentially detect such structural changes from observed demand, making only minimal assumptions on the underlying demand model.