A Study on Order Forecasting Model for Graphic Card Industry

• Main Authors: Liu,Chia-Chen, 劉家陳
• Other Authors: Chen,Chun-Hsien
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/20607243936983734108

碩士 === 長庚大學 === 資訊管理研究所 === 94 === With the rapid improvement on IT industry, the market value percentage of integrated graphic processors is more than 56% in the market of graphic processors. Thus the graphic card industry gradually enters an era of low gross profit. Manufacturers of graphic card are beginning to face serious competition.It is believed that only effective management forecast as well as stock control can effectively emerge as winner in this market of drastic competition. This research is focused on the forecast of short-term order. This short-term order will eventually affect the stock management control. Low gross profit and ineffective stock control will gradually lead to a loss of profit. Accurate forecast on short term order is an important reason for manufacturer in making profit. This thesis evaluates three methods for order forecast ,including average moving method, exponential smoothing method, and back propagation neural networks through careful studies on past sales order.The historical data of sales order is form different categories of graphic cards. In the research,MAPE(Mean Absolute Percentage Error) is used for performance evaluation. The study shows that the back propagation neural network works best and the average moving method works worst among these three methods.