非线性时间序列周期分析的R/S分析方法及其应用

许多时间序列 ,例如资本数据等经济类时间序列 ,是由众多错综复杂因素共同作用的结果 ,存在着种种线性和非线性作用机制 ,频谱分析及其种种变形不应该是这些时间序列周期分析的合适工具 ,R/S分析因为不象频谱分析那样有正弦或余弦的假设 ,因而具有明显的优势 .通过对上证指数的 R/S分析 ,发现上证指数具有长程正相关和大约 5个月一个周期的特点 .

R/S Analysis for Finding Non-linear Time Series' Cycle and Its Application

Some time series, such as capital data series and other economic data time series, come from interaction of large number of intricate factors and possess linear/non-linear mechanism. The conventional spectral analysis and other similar statistical tests would be inappropriate tools for cycle analysis on such time series. Here, the superiority of R/S analysis appear owing to lacking hypothesis of sine and cosine. The result of R/S analysis shows that Shanghai composite index is characterized by a persistent process for periods up to about 5-month.