標題: | 以ARIMA季節相乘模式預測汽車售後服務進廠台數之研究-以裕隆日產體系為例- Forecasting for the Amount of Aftersales Repair Orders of the Automobile Industry by Using a Multiplicative Seasonal ARIMA Model - A Case Study of YULON-NISSAN Group - |
作者: | 沈志陽 Chih-Yang Shen 姜齊 Chi Chiang 管理學院管理科學學程 |
關鍵字: | 預測;ARIMA;汽車產業;forecasting;ARIMA;automobile industry |
公開日期: | 2007 |
摘要: | 台灣汽車產業每年產值超過新台幣三仟億元,對國內經濟發展擔任非常重要角色,但自2000年加入WTO後新車市場銷售競爭激烈,近兩三年因油價逐年高漲及卡債風波影響,新車掛牌數量成長率已連續十季下滑,新車製造與銷售相關廠家如製造廠、總代理及經銷公司在新車部份僅能維持小幅獲利甚至虧損,面臨汽車產業嚴峻之挑戰。
在新車市場經營困境之下,汽車售後服務成為車廠及經銷公司舉足輕重之利潤來源,進廠台數預測準確性重大影響公司服務營收、新車及售後服務行銷活動、成本費用預算,進而影響整體獲利及獲利率,故適當之進廠台數預測模式益形重要。
台灣汽車產業對售後服務進廠台數之預測以專家法為主,缺少信度與效度之檢證,實難提供經營者具理論基礎及高度準確之預測數值。本研究應用時間序列之ARIMA季節相乘模式,以裕隆日產體系售後服務進廠台數為例,根據1999年至2006年進廠台數之數據,建立未來進廠台數所呈現之ARIMA預測模式,並依預測模式計算年度之季節指數,將2007年預測值與實際值比較分析,得出最適預測模式。
研究結果建立最適ARIMA預測模式為ARIMA(0,1,1)*(2,1,0)12,顯示ARIMA季節性相乘模式適合作為汽車產業售後服務進廠台數預測之參考模式。 The output value of Taiwan automobile industry is over 300 billion NT dollars every year. It plays a very important role for domestic economic development. When Taiwan entering WTO in 2000, new vehicle market competition gets harder and harder. The growth percentage of new vehicles has been decreasing ten quarters continuously because of the price raising of petroleum and disturbance of credit cards. Taiwan automobile industry is facing the rigorous challenge. In this predicament of new vehicle market, the profit from aftersales service becomes the major source. The prediction accuracy of the amounts of aftersales repair orders will affect service income, the marketing activities and budget. And it also influences the profit and its rate. So the appropriate prediction model of the amount of aftersales repair orders is even more significant. The main method of predicting the amounts of aftersales retention is specialist method. It is short of reliability and validity. So it is hardly to give the executives the reliable prediction. This study applies the multiplicative seasonal ARIMA model using the data of YULON-NISSAN Group from 1999 to 2006 to establish a prediction model, and calculate the season indexes. By comparing the forecasted and actual amounts of 2007, the proposed model is recommended. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#GT009562516 http://hdl.handle.net/11536/39786 |
Appears in Collections: | Thesis |