Prediction Of Change Points Based On RRINARCH Model

There are many change points in long time series.These change points in chronological order can be regarded as an integer value time series based on change points observation.In this paper,LRSM is used to estimate the change points of time series.We assume that the change points is only affected by fragment lengths.In this paper?we use RRINARCH model to fit the fragment lengths,and use the fitted RRINARCH model to predict the new change point of time series.Furthermore,the prediction interval of the new change point is obtained by using the bootstrap method.Simulation experiments show that our new method gives very accurate prediction and interval estimation of the change points.Finally,we apply the new method to real time series data.