基于时间序列异常检测分析的方法

Method of Anomaly Detection and Analysis Based On Time Series

不依赖于模型,基于累计变化量来实现异常点的检测,而无法检测出成片的异常点,并且很容易把正常点视为异常点,为了解决上述方法中所存在的问题,本文在所给方法的基础上重新定义了累计变化量,引入了推移算子,异常类型指示变量和异常惩罚量,并定义了两类异常类型,一种叫做高位异常,一种叫做低位异常,然后重新定义了异常点模型,引入了然后用2004年到2009年的沪市股票数据来进行数值实验,并对结果进行了对比证明了本文所给方法的有效性。

About not dependent on the model and is relatively simple and easy to implement about the methods of the time series anomaly detection, but it can not detected a piece of outliers, and it is easy to make normal points as outliers. In order to solve the problems, on the basis of the method given, this paper redefines the cumulative change and introduces the transition operator, one indicator variable and unusual punishment are introduced in this paper and two exception types are defined, one is called the high anomaly, another is called the low abnormal. The effectiveness of the method given in this article is proved by using the data from Shanghai Stock Market between 2004 and 2009 be proved through numerical experiments. The results are compared to prove the effectiveness of the method presented in this paper.