透過您的圖書館登入
IP:18.191.157.186
  • 學位論文

連續型自迴歸時間模式之貝氏分析應用

On Continuous Time Threshold Autoregressive Model : a Bayesian Application

指導教授 : 林余昭
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


對於連續型非線性的時間序列模型在時間序列的分析上有著相當重要的應用,但通常其概似函數卻無法獲得, 因此在分析上就會比線性的模型困難許多。 一些文獻當中, 例如: Tong (1983), Tsay (1986, 1989),Tong and Yeung (1991), Brockwell, Liu and Tweedie (1992),對於非線性的模型已有所討論。 在此篇論文當中, 我們根據 Roberts and Stramer (2001) 及 Lin (2003) 利用貝氏統計方法對連續型時間門檻自迴歸過程 (continuous time threshold autoregressive model) 做分析。 在應用上, 我們套用的是 S&P 500 的資料,其為一個從 1994 年 1 月 3 日為期約兩年的股市交易收盤指數。 在先前的文獻當中,都是對資料時間間隔為相等的 (equally spaced data) 來做探討,但在實務上,我們能夠得到的資料往往與等間隔的時間間隔此項假設不符, 因此本文將改正這樣的假設,針對整個過程中造成不等的時間間隔的遺失資料作處理。

並列摘要


The non-linear continuous time models have important applications in time series analysis. However, their likelihood functions are usually not available. As a result, the analysis is trickier than that of their discrete time conunterparts. In this study, we use the Bayesian method to analyze the continuous time threshold autoregressive models. This approach is based on Roberts and Stramer (2001) and Lin (2003). In the applications to financial data S&P 500, we assume the daily data are not equally spaced since the stock market only opens on weekdays and we treat the paths between observed data are missing.

參考文獻


of stationary threshold autogressive moving-average processes. J.
Time Ser. Anal. 13, 95-107.
[2] Brockwell, P.J. (1994), On continuous-time threshold ARMA processes.
J. Statist. Planning Inf.,39, 291–303.
[3] Brockwell, P.J. (2001), Continuous-time ARMA Processes, In

延伸閱讀