Determining Valid Underlying Retail Demand Rates for Proposed Prediction Model of Slow Moving Merchandise

Predicting the future demand rate of retail merchandise that has not experienced consumer demand over a specific interval of time is a challenging problem that has not received sufficient attention in the literature. A methodology is proposed to estimate the future demand rate of “slow” moving products. The distribution of the estimator used in constructing the proposed prediction intervals has previously been illustrated to be skewed for certain parameters. This paper investigates the range of underlying demand rates that allow the proposed prediction intervals to have valid confidence levels. The interval of time selected to observe the demand rate of slow moving merchandise is selected to be 100 time periods to approximate 3 quarters with the time periods representing days. A simulation study reveals that two-sided prediction intervals are valid over a wider set of parameter values for the underlying demand rates of the products than one-sided prediction intervals. Prediction intervals are investigated at the 90%, 95%, and 99% confidence levels. CONSTRUCTION OF PROPOSED PREDICTION INTERVALS FOR FUTURE LOW DEMAND RATES This research is motivated by a renewed interest in forecasting demand for retail products that are often characterized by infrequent transactions (Willemain, Smart, Shockor, & DeSautels, 1994; Johnston & Boylan, 1996; Syntetos & Boylan, 2001). Inventory, or a stock of goods, is a necessary part of business and comes in many forms. At the retail level, its main function is to allow the firm to meet expected customer demand and prevent shortages. However, holding inventory presents real costs to companies and managers are under pressure to reduce inventories at all levels to reduce the related costs (Masters, 1993). Techniques to minimize inventory costs, while still providing acceptable levels of customer satisfaction, have been widely studied, with the exception of products with low demand rates (Hollier, Mak, & Lai, 2002). To cope with the