Research on network data forecast system based on non-linear time series

Network data forecast system which is one of the main studies in system modeling plays an important role in system. Because the network data is not stationary and there are unpredictable factors, nonlinear time series modeling method should be adopted to analyze and forecast it. Based on the analysis of the data network, the ARIMA model is established. When the order of the prediction model is determined and parameter estimation is done, forecasts of network data flow under the conditions of different forecast step are given, and comparison simulation experiments are carried out. Simulation results show that, the model´s forecast error is around 4% in predicting the smaller step, so it has good prediction accuracy and provide a solid foundation for network data flow forecast, anomaly detection and network load forecast.