在目前统计预测中，存在着非平稳序列分析效果差、多步预测误差较大、缺

乏系统的软件实现等问题。本文针对该类问题进行研究，提出了NARIMA方法，

该方法以ARIMA模型为基础，有效结合了游程平稳检验方法、差分平稳处理方法

、线性最小方差预测算法等，解决了传统统计预测方法中存在的上述问题

-In the current statistical forecast, there is a differential effect of non-stationary sequence analysis, multi-step prediction error, lack of system software implementation. For this kind of problem, NARIMA method, the ARIMA model, the effective combination of a smooth run-length test, the differential steady approach, the linear minimum variance prediction algorithm to solve the traditional statistical forecasting methods exist the above problems