台灣矽晶圓出口值預測不同模式比較研究

Comparison of Export Value Forecasting Models for Silicon Wafers in Taiwan

施明豪(Ming-Aur Shih)

《商業現代化學刊》 4卷4期 (2008/09) Pp. 135-146

https://doi.org/10.6132/JCM.2008.4.4.09

摘要

隨著國際原油價格創下新高及環保意識抬頭，世界各國紛紛投入研發製造再生能源及生質能源，其中廠商佈局乙太陽能爲最積極，太陽能廠商每年投入不少的金額去擴廠及佈局，深怕在太陽能市場被淘汰，太陽能上、中、下游，又以上游矽晶圓缺料爲最嚴重，本研究的目的是爲了避免矽晶圓生產不足或過剩，以降低業者的損失。 本論文主要以2004年1月至2007年12月台灣矽晶圓的出口值，以灰色理論（逐月計算）、灰色理論（無逐月計算）、簡單線性迴歸分析、單指數平滑法四種預測方法，來建立矽晶圓之出口預測模式，並利用絕對百分比誤差(MAPE)誤差指標，選取出其指標數值最小，做爲選取最佳預測模式，進行預測2008年1月之出口值，以提供矽晶圓未來挑戰的因應策略，提供政府或相關單位作參考。

關鍵字

預測 ； 矽晶圓 ； 灰色理論 ； 迴歸分析 ； 指數平滑法

並列摘要

Affected by surging oil prices and growing environmental awareness, nations worldwide have been investing in R&D and manufacturing of recycling energy and biomass energy. Among which, manufacturers have been proactively engaged in solar cell development. Solar cell manufacturers have invested heavily to expand production and deployment, afraid of being eliminated from market share. In the upstream, midstream and downstream of solar cell development, the silicon wafer, in its upstream mode, suffers from serious material shortages. This research aims to address the problems associated with insufficient production (and, surpluses of silicon wafers), in order to decrease the losses to manufacturers.
The research focuses on export value of silicon wafers from January 2004 to December 2007, uses the Grey Theory computed every month (and the Grey theory applied less frequently), simple linear regression analysis, and single exponential smoothing, in order to establish an export forecasting model for silicon wafers. Additionally, this research adopts MAPE as the best forecasting model, to forecast its export value for January 2008. It is hoped that the results of this research can lead to silicon wafer production strategies, in order to better face future challenges, and to serve as a reference for government and the related industry.

並列關鍵字

forecasting ； silicon wafer ； Grey Theory ； Regression Analysis ； exponential smoothing