透過您的圖書館登入
IP:3.129.211.87
  • 學位論文

時間數列與灰預測模型的比較-以台灣自行車成車出口產值為例

The Comparison Analysis of Time Series Models And Grey Forecast-An Example of the Production Value Forecast of Bicycle Industry in Taiwan

指導教授 : 杜震華
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


台灣是自行車王國,在兩次石油危機,自行車出口產值皆大幅成長。面對2004年以來不斷高漲的石油價格,台灣自行車出口的產值又有挑戰歷史新高的機會。台灣的自行車歷經四個發展階段,分別是裝配生產階段、擴大出口階段、產業轉型升級階段與國際化階段,從傳統的OEM代工接單過渡到ODM設計代工再過渡到推出自有品牌IBD,面對國際市場的競爭與逐漸高漲的勞動成本,自行車產業不但要在產品上不斷推陳出新,更要能預測未來市場以因應榮枯所帶來的衝擊。 本研究是比較數個線性時間序列模型與非線性灰預測模型,求出最適的自行車出口產值預測模型。在時間序列的假設裡,必須滿足穩態型時間數列的要求,或利用差分使數列成穩態型,再進行預測。灰預測模型在程序上就相對簡化,只要樣本數大於4,即可用以預測。在出口產值資料上,以經濟部國貿局與臺灣區車輛工業同業公會1997年1月到2007年9月的自行車成車出口產值為選取樣本,並以平均絕對值誤差百分比、泰爾係數與均方根誤差百分比為模型預測評判標準,選出最適預測模型以預測2007年10月以後的出口產值。 本研究發現時間序列之季節性與非季節性自我相關整合移動平均模型預測模型的表現最佳,因其預測在平均絕對值誤差百分比、泰爾係數與均方根誤差百分比都是誤差最低的,本研究並以此模型做為2007年10月以後台灣自行車成車出口產值之最適預測模型。

並列摘要


Taiwan is the kingdom of bicycles. During the period of the Two-Crisis of Petroleum, the export product value of the Taiwan bicycle grew rapidly. When facing the increasing petroleum price after 2004, the Taiwan bicycle’s export product value seems to have a chance to brush a new history record. The development of the Taiwan bicycle industry can be distinguish four periods:the resembling period, the export extension period, the transforming period, and the internationalize period. The product way from traditional Original Equipment Manufacturer(OEM)transited to Original Design Manufacturer(ODM)and finally to Individual Bicycle Dealer(IBD). Facing the competition among the international countries and the gradually increasing wage cost, the Taiwan bicycle industry needs not only to weed through the old to bring forth the new; but also to forecast the shock whether it is positive or negative. This research aims to compare the performance of linear time series forecasting model and non-linear grey forecasting model GM(1,1)to obtain the optima forecasting model and to predict the bicycle export product value. Under the assumption of the time series, the series needs to meet stationary status or by difference to be, then applying to forecast the next time. However, the Grey forecasting model seems to simplify in process, which needs only four observations to be done. The bicycle export product value data excerpt from the Bureau of Foreign Trade and the Taiwan Vehicle Craft Union published on the web, from January 1997 to September 2007. After selected by Mean Absolute Percentage Error(MAPE)、THEIL U and Root Mean Squared Percentage Error(RMSPE)methods, we use the best model to forecast the product value after October 2007. This research finds the SARIMA has the best performance no matter under MAPE、THEIL U or RMSPE test, because of the relative low expect error on above, and we use the model to forecast the future value after October 2007.

參考文獻


葉斯水(2005),《台灣自行車產業發展之回顧與前瞻》,元智大學管理研究所碩士論文。
陳俊瑺(2005),《中國大陸石油需求之實證分析》,國立台灣大學國家發展研究所碩士論文。
于乃茂(2006),《台灣自行車產業市場結構、行為與績效之研究》,國立台灣大學經濟研究所碩士論文。
俞慧芸(2004),《台灣自行車產業轉型的歷史考察(1970-1990)》,國立清華大學社會學研究所台灣社會學會年會。
林俊宏(2004),<台北都會區捷運站腳踏車停車需求之研究>,《都市交通季刊》,19(3),頁16~31。

被引用紀錄


陳燕玲(2014)。應用灰色理論於黃金價格之預測〔碩士論文,國立臺灣大學〕。華藝線上圖書館。https://doi.org/10.6342/NTU.2014.11069

延伸閱讀