An Integrated Forecasting DSS Architecture in Supply Chain Management

Tien-You Wang
Tainan University of Technology, Taiwan R.O.C

Din-Horng Yeh
National Chung Cheng University, Taiwan R.O.C

In a competing market environment, supply chain management (SCM) has been critical for companies to survive. Demand planning plays an important role in SCM, for it provides accurate demand forecasts which may achieve customer satisfaction by offering benefits such as low inventory level, short lead time, efficient resource allocation, and quick response. To obtain more accurate forecasts, this study presents a web-based DSS architecture and its forecasting core. The forecasting core, named Panel Function, contains three modules: Segmentation Module, Forecasting Module, and Coordination Module. Segmentation Module employs data mining technology to categorize customers with different characteristics into three segments: Loyal Customer Segment, Potential Customer Segment, and Switcher Segment. Based on the three segments, Forecasting Module employs different forecasting and analysis technologies to estimate an integrated forecast: time-series forecasting to capture the loyal customer demand trend, Bayesian inference to estimate the predicted value of switcher purchase quantity, and questionnaire analysis and brand choice models to unearth potential customers. An integration function then synthesizes the results from these three processes to obtain the integrated forecast. Coordination Module then takes this integrated forecast as the base of distribution planning, and provides a minimal system-wide total cost solution for all parties in the supply chain. As a whole, this DSS architecture is anticipated to provide an efficient mechanism for collaborative demand planning, and help create the maximum profit for the supply chain.

Download Full Paper