稳健AR模型的构建及其在金融时序中的应用

发布时间：2018-03-20 18:05

本文选题：AR模型　切入点：稳健统计量　出处：《统计与信息论坛》2017年05期 　论文类型：期刊论文

【摘要】：时间序列自回归AR模型在建模过程中易受离群值的影响,导致计算结果与实际不相符。针对这一现象,运用FQn统计量对传统自相关函数进行改进,构建出自回归AR模型的稳健估计算法,以克服离群值的影响,并对此方法进行了模拟和实证分析。模拟和实证分析均表明:当时序数据中不存在离群值时,传统估计方法与稳健估计方法得到的结果基本保持一致;当数据中存在离群值时,运用传统估计方法得到的结果出现较大变化,而运用稳健估计方法得到的结果基本不变.这说明相对于传统估计方法,稳健估计方法能有效抵抗离群值的影响,具有良好的抗干扰性和高抗差性。
[Abstract]:The autoregressive AR model of time series is easily affected by outliers in the process of modeling, which leads to the inconsistency between the calculated results and the actual results. In view of this phenomenon, the traditional autocorrelation function is improved by using FQn statistics. A robust estimation algorithm based on regression AR model is constructed to overcome the influence of outliers. The simulation and empirical analysis show that: when outliers do not exist in time series data, The results obtained by the traditional estimation method are basically consistent with those obtained by the robust estimation method, and when there are outliers in the data, the results obtained by the traditional estimation methods vary greatly. The results obtained by the robust estimation method are basically unchanged, which shows that the robust estimation method can effectively resist the influence of outliers, and has good anti-interference and high robustness compared with the traditional estimation method.
【作者单位】：广东财经大学统计与数学学院;
【基金】：国家社会科学基金项目《稳健统计过程控制的大数据分析方法研究》(16BTJ035) 广东省自然科学基金项目《稳健过程控制图的构建及评价方法研究》(2016A030313108)
【分类号】：F224;F830

【相似文献】