Abstract:
|
We are interested in robust strategies which could control the Cesaro loss to Bayes Envelope in the sequential plays, for example, costumers compliance to certain services. We focus on play against the random past strategy (PRP), which is derived from the Hannan's idea but we use a different approach. The PRP strategy was first proposed by Gilliland and Jung (2006), in the context of classical matching binary bits problem. In our previous studies, we defined and studies PRP strategy in higher dimensional matching binary bits game,and investigate k-extended Bayes envelope problem in an infinite state decision. However, all these work were based on symmetric repeated plays, and the risk evaluation over the sequential prediction was also based on symmetric pay-offs. We also explore PRP strategy in non-symmetric sequential predictions. In this paper, we would like to explore PRP strategy's asymptotic properties in more complicated sequential prediction scenario, through simulations of real Business data.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.