基于时间序列干预分析模型的我国铁路客运趋势预测研究
A Study on Trend Prediction of China’s Railway Passenger Transport Based on Time Series Intervention Analysis Model

为评估突发事件对我国铁路客运量的趋势影响，本文提出在识别时间序列离群值和最优ARIMA模型基础上，对干预影响序列进行基于脉冲函数或阶梯函数的虚拟变量针对性设计，确定干预过程的传递函数，与最优ARIMA模型结合，建立干预预测模型。文章选取2015年1月至2021年12月的全国铁路客运量月度序列数据建立模型，通过研究发现：疫情的影响导致原本具有周期性、增长趋势的客流序列产生离群值，并且干预发生后对滞后期仍有影响；干预分析模型拟合程度较好，预测符合实际走向，模型的均方根误差(RMAE)为2856.15949，总体上看精确度较高。
To assess the impact of emergency events on the trend of China’s railway passenger volume, this paper proposes a targeted design of dummy variables for the intervention impact sequence based on impulse functions or step functions, after identifying outliers in the time series and selecting the optimal ARIMA model. The study determines the transfer function of the intervention process and combines it with the optimal ARIMA model to establish an intervention prediction model. Using monthly series data of national railway passenger volume from January 2015 to December 2021, the research finds that the impact of the pandemic has resulted in outliers in the originally periodic and growing passenger flow sequence, and the intervention still has an effect on subsequent periods. The intervention analysis model demonstrates a good degree of fit, with predictions aligning with actual trends. The root mean square error (RMAE) of the model is 2856.15949, indicating a high overall accuracy.