Study On Method And Application Of Multiple Change-points Identification Based On Non-parametric

Statistical process control (SPC) is an important method of monitoring theprocess. Control charts are effective and widely used as process monitoring tools. Theconstruction of control charts can be divided into two stages: Retrospective phase andmonitoring phase. Generally, current study assume that the underlying data comefrom a single run mode, but in real life there is usually more than one operating mode.In order to achieve the second stage effectively, there is a need to accurately identifythe transformation point in the first stage of the process and extract the normal processof the operating state.In the retrospective phase, the issue of identifying the change point is hot. At first,in the case of a single variable, univariate point is the research object. Then the scopeof the study expands to multiple change points and multivariate. However, the studyof multiple change points is still incomplete at present, what’s more, the process isusually assumed obey to a known parameter distribution, so the current study has ahighly dependence of specific distribution. In view of this, this article proposes amultiple change points identification process without any assumption of the processbased on non-parametric statistical method. Specifically, it is a study to a variety ofchanges in the type of drift under the KS test and the Mann-Whitney U test.The effects of unexpected events can’t be ignored. What’s more, stock marketrisk has a close relationship with the public life. According to a large number ofstudies, the distribution of the stock market is not only subject to the normaldistribution, at the same time, it can’t be accurately fitted by a certain kind ofdistribution. Substantially, the distribution of the stock market has a fat tail. So, thecurrent parametric method does not apply to the change point identification of stockmarket. Nonparametric change point identification method proposed in this paper canbe good to make up for the deficiencies of the above studies. In order to verify thevalidity of the identification method, this article takes Shanghai A Share as anexample.