標題: | 檢定兩個自迴歸移動平均模型或兩個隨機係數自迴歸模型的相等性 Testing equality of two ARMA models or two random coefficient autoregressive models |
作者: | 鍾興潔 Jong, Sing-Jie 王秀瑛 Wang, Hsiu-ying 統計學研究所 |
關鍵字: | 時間序列;自迴歸移動平均模型;隨機係數自迴歸模型;time series;ARMA model;RCA model |
公開日期: | 2011 |
摘要: | 時間序列分析是一套動態數據處理的統計方法。基於隨機過程及數理統計理論,分析變數的產生與變數間之動態關係,進而能檢定經濟理論或對變數進行預測,以用於解決實際問題,其中,又以Box-Jenkins的自迴歸移動平均模型之分析方法最廣為被大家所使用。此外,理論的發展與推廣,時間序列模型至今已發展到相當複雜的程度,隨機係數自迴歸模型(Random Coefficient Autoregressive model;RCA)便是一個值得深入研究的主題,在這篇論文裡,我們提出檢定兩個自迴歸移動平均模型相等性的方法以及檢定兩個隨機係數自迴歸模型相等性的方法,並且在我們的模擬結果中顯示,我們的方法確實能使該檢定的型一錯誤達到我們設立的顯著水準。我們應用這個分析方法在實際的公司營收資料上,在我們所分析的三間公司中,顯示出有兩間公司的營收在我們所配適的模型上有顯著的差異。 This thesis addresses the issues of testing equality of two time series models. Testing procedures for testing the equality of two ARMA models or two random coefficient autoregressive (RCA) models are proposed. For testing the equality of two ARMA models, we based on the maximum likelihood estimators to establish a testing procedure. For testing equality of two RCA models, an empirical likelihood method is developed. The proposed methods have been demonstrated to have good properties and are shown to have good performance through simulation studies. Also, the testing procedure for testing the equality of two ARMA models is illustrated through an analysis of three companies’ monthly sales. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#GT079926520 http://hdl.handle.net/11536/49929 |
顯示於類別: | 畢業論文 |