论文部分内容阅读
替代数据检验法是检验时间序列中是否存在确定性非线性成分的重要统计方法.通过研究差分和数据平滑运算对替代数据检验方法的影响,指出常用的线性滤波等数据预处理步骤破坏了序列的静态性质,从而会导致对零假设的错误拒绝.因此,建议应该直接利用原始时间序列而非应用了差分等非静态滤波运算后的时间序列生成替代数据,再进行假设检验,以免造成对零假设的错误拒绝.
The alternative data test is an important statistical method to test the existence of deterministic non-linear components in the time series.Through the study of the effects of differential and data smoothing on the alternative data testing methods, it is pointed out that the commonly used linear filtering and other data preprocessing steps destroy the sequence Static nature, which will lead to false rejection of null hypothesis.Therefore, it is suggested that alternative data should be generated by using the original time series rather than the time series after non-static filtering operations such as the difference, and then the hypothesis test should be conducted to avoid the null hypothesis The wrong refusal.