The Optimal Decision Model for Multi-Period Stochastic Demand Problem：A Dynamic Bayesian Approach

• Main Authors: Jhang, Houng-Bing, 張鴻彬
• Other Authors: Wu, Ji-Cheng
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/69455950608209762406

碩士 === 國立屏東科技大學 === 工業管理系所 === 98 === The traditional newsboy problem is to decide the stock quantity of an item when there is a single purchasing opportunity before the start of the selling period and the demand for the item is random. However, the most order way and quantity for perishable goods are multi-period and dynamic decision in the market. On the other hand, a lot of research work has been done in the newsboy problem to seek for the optimal order quantity, but most studies have focused on the classical statistical methods. There has been relatively little research conducted on the Bayesian statistical framework. In fact, retailers often can give the short, medium even long-term forecast for the state of demand according to the historic sales record and their professional judgments. Therefore, we adopt the Bayesian point of view to construct the dynamic Bayesian model for multi-period newsboy problem that combine the prior knowledge of the decision makers and the actual sales information. First assume that the consumer population, sales channel and other conditions of sale have not changed much in finite-horizon planning. In other words, the state of demand is stationary process. Thus, we use the constant mean model to fit the relationship between the demand of the goods and the state parameter. In the model, the observation equation is to describe how the demand relies on current state parameter and the state equation is to describe the former one and the next changes in parameter. In case of over-stock and under-stock, we use the normal distribution theory and two moments approximation to adjust the posterior to become prior distribution, respectively. In this way, the expected total cost function can be obtained by iteration on the ground. Decision-makers will be able to more accurately grasp the changing demand for goods, and to develop dynamic adjustment of optimal inventory policies. Finally, we propose an example to illustrate the application of the Bayesian methods for multi-period newsboy problem, as well as the optimal order quantity of iterative calculations.