Extension on Cut-off Rules of Sequential Observation and Selection Problem

Abstract

Abstract Cut-off rules specialized in sequential observation and selection problems (SOSP). A class of SOSP was studied, in which a decision maker (DM) face multi-times selection case with the objective of minimizing the average rank of all selected alternatives. In modeling this class of problems, the article firstly present the optimal cut-off rules that use maximum, before cut-off, as benchmark, and then report results from simulation experiment. Furthermore, the article proposes that to fall benchmark is more efficient than to decrease cut-off value for improving the average payoff. Thus, the optimal cut-off rule was revised by replacing their maximum benchmark with average benchmark based on the risk aversion and extreme aversion. The results from simulation experiment indicate that the revised cut-off rule has better average payoffs. At last, a proof process that the 20 percent is the approximate optimal ratio to estimate the benchmark brings this article into a close.