把时间序列分析的方法和理论引入业务监控中。本文的创新点

在于利用AR州人模型建模前，通过孤立点和变点检测对数据做预处理，

先用控制图法去除孤立点，然后用变点检测的方法定位变点，最后对变

点前后的时间序列分段预测，从而提高了预测的准确度和可信度

-Time series analysis methods and theoretical introduction of the service monitoring. Innovation of this paper is to use the AR-state modeling, outlier and change point detection on the data pre-processing, to use the control chart method to remove isolated points, then change point detection method to locate the transition point, and finally change the sequence of time points before and after segment forecasts, thereby improving the accuracy and credibility of the forecast