یک روش بیزین برای تعیین ارزش اطلاعات در مسئله پسر روزنامه فروش
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|22649||2008||12 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : International Journal of Production Economics, Volume 112, Issue 1, March 2008, Pages 391–402
The potential benefits of demand forecasting in a newsboy problem are to decrease the risk of overstocking or shortage, but forecasting is usually not free. In this paper, a model to help the decision-maker in a newsboy problem to assess the value of information is presented. First, two cases on available demand information are considered to develop an upper bound of the cost that the decision-maker would spend on forecasting. Then, a Bayesian approach to forecasting is proposed and EVAI, the expected value of additional information, is computed to help the decision-maker in deciding whether to use the extra information. Finally, the relationship between the EVAI and model parameters is discussed.
In business operations, forecasts are the basis for budgeting, planning capacity, sales, production and inventory, purchasing, personnel, distribution, financing, and more. Demand forecasting is particularly important for the newsboy problem since the shelf life of goods is limited. In general, forecasting methods can be divided into two types: qualitative and quantitative. When assessing forecasting methods, accuracy is not the only criterion to be considered; the cost of forecasting is also an important factor. In this paper, a model to help the decision-maker in a newsboy problem to assess the value of information is presented and a Bayesian approach to forecasting is proposed. Then, some useful managerial insights about the value of information are derived to guide the decision-maker when forecasting demand. The rest of this paper is organized as follows. Section 2 presents an overview of the most relevant literature. Section 3 proposes our models and its analysis. Section 4 reports our numerical studies. Section 5 concludes the paper and points out possible future research directions.
نتیجه گیری انگلیسی
In this paper, we develop a model for the decision-maker in a newsboy problem to assess the value of demand forecasting. Firstly, we derive an upper bound of the cost that the decision-maker would spend on forecasting—denoted as EVPI. Thus, the users in practice can allocate the budget of forecasting by the upper bound. This is very important since demand forecasting is the basis of business planning and insufficient budget may cause inaccurate forecast affecting the performance of business planning. Then, we consider a sequential decision process. In this process, the decision-maker needs to decide whether to use a Bayesian forecasting method, then to decide the order quantity. In our model, we assume that the forecast is an unbiased and leading estimator of demand, and is used to revise prior demand distribution to obtain posterior distribution. Furthermore, the expected value of additional information from the Bayesian forecasting method, EVAI, is derived. If EVAI is larger than the cost of forecasting, the decision-maker may consider adopting the Bayesian forecasting method; otherwise, the decision-maker should use prior information only. Other important results are as follows: (1) There exists a threshold value of forecast. If forecast is greater than the threshold value, the decision-maker will issue an order. (2) Having extra information (using the Bayesian forecasting method), the decision-maker is more sensitive to the increase in a positive model parameter and insensitive (less vulnerable) to the change in a negative parameter. In our model, the positive parameters are unit underage cost and prior mean; the negative parameters are unit overage cost and setup cost. Finally, future research directions may be as follows: (1) Add a constraint of limited capacity into the models. (2) Relax the assumption of unbiasedness, equal variance and normality. (3) Use a forecasting method different from the Bayesian method. (4) Extend to multi-item or multi-period models.