Prediction Intervals for an Unknown Mean Gaussian AR(1) Process Following Unit Root Tests

摘要：

This paper presents two new one-step-ahead prediction intervals for an unknown mean Gaussian AR(1)process. We propose the simple prediction interval based on the residuals model, PIa, and the prediction interval following the unit root tests, PIf. The unit root tests applied in this paper are the Dickey-Fuller unit root test, the Phillips-Perron unit root test, the weighted symmetric unit root test, and the Elliott-Rothenberg-Stock unit root test. The coverage probabilities of all prediction intervals are derived. The performance of the proposed prediction intervals is assessed through Monte Carlo simulation studies. Simulation results have shown that all prediction intervals have minimum coverage probabilities 0.95 for all the autoregressive parameter values. Moreover, the expected lengths of prediction intervals PIf are shorter than that of a prediction interval PIa when the autoregressive parameter value is close to one.

关键词： AR(1) Unit Root Test Prediction Interval Preliminary test Residual Model

作者: Sa-aat Niwitpong Wararit Panichkitkosolkul

作者单位: Department of Applied Statistics King Mongkuts University of Technology North Bangkok Bangkok, Thailand

会议类型: 国际会议

会议名称: The 10th International Conference on Intelligent Technologies(第十届智慧科技国际会议 InTech09)

会议地点: 桂林

会议语种:英文

页码: 221-229

在线出版日期: 2009-12-12（万方平台首次上网日期，不代表论文的发表时间）