Stochastic Shortest Path and its applications to golf
Supervisor Name : Gautier Stauffer Doctoral School : MSTII Start Date : 2015 1th September Financing - Context - Partnerships: Research allowanceof the Graduate School/MinisterialExchange /AOthematicdoctoralContractsUJF Positioning and Challenges :
The goal of the project is to build mathematical models and algorithms that exploit the huge mass of data, provided by ShotLink (the company collecting data for the Professional Golf Association in the United States) and the new mainstream sensors that can collect ball trajectories (such as trackman or flightscope), to create decision support tools for golf. We propose to first develop tools that exploit players’ statistics in order to optimize their game depending on the golf course and then to use those results at different level in order, for instance, to: help players improve their performances, forecast the winner of the next tournament, understand how to modify the course in order to make the competition more balanced or harder, define better course ranking systems, etc…
« Natural » methods to solve this kind of issues are based on stochastic programming and in particular on Markov Decision Process and stochastic shortest path. Those tools have proved powerful in the past when optimizing production systems and we have a great expertise in this area in the lab. We propose to capitalize on this know-how to develop innovative solutions to those new emerging problems. The size of the instances might have a big impact on the applicability of the standard methods and it might thus be necessary to extend the current tools and/or to develop new methods. This project will therefore focus both on theory and application.