考慮需求不確定性之動態訂價與資源分配問題

Abstract

In this study, we investigate a pricing and capacity allocation problem with fixed capacity and uncertain customer demands to maximize the expected revenue over a finite selling horizon. We assume that each period’s customer demand depends on selling price and the time period. At the beginning of each period, the decision maker faces two problems simultaneously: (1) how to allocate the remaining capacity among the remaining selling periods; and (2) how to price the capacity allocated on this period. The goal is to maximize expected total revenue. First, we develop two algorithms, CAAMSP-a and CAAMSP-b, to allocate the capacity among selling periods with each period’s predetermined price. Second, we propose a probabilistic dynamic programming model to determine each period’s price, with a given capacity allocation policy. Finally, we propose two additional algorithms, OCAMSP-a and OCAMSP-b, to find a near optimal capacity allocation policy and each period’s price simultaneously for maximizing expected total revenue over a finite selling horizon.

In this study, we investigate a pricing and capacity allocation problem with fixed capacity and uncertain customer demands to maximize the expected revenue over a finite selling horizon. We assume that each period’s customer demand depends on selling price and the time period. At the beginning of each period, the decision maker faces two problems simultaneously: (1) how to allocate the remaining capacity among the remaining selling periods; and (2) how to price the capacity allocated on this period. The goal is to maximize expected total revenue. First, we develop two algorithms, CAAMSP-a and CAAMSP-b, to allocate the capacity among selling periods with each period’s predetermined price. Second, we propose a probabilistic dynamic programming model to determine each period’s price, with a given capacity allocation policy. Finally, we propose two additional algorithms, OCAMSP-a and OCAMSP-b, to find a near optimal capacity allocation policy and each period’s price simultaneously for maximizing expected total revenue over a finite selling horizon.