后疫情时期秦皇岛站铁路客运量的统计建模与预测效果分析

Statistical Modeling and Prediction Analysis of Passenger Volume of Qinhuangdao Railway Station in Post-Epidemic Period

摘要: 后疫情时期，铁路客运需求发生巨大改变，突如其来的疫情中断其逐年增加的趋势。建立合适的模型预测客运量，对铁路运营工作的顺利展开起着至关重要的作用。论文运用2020年1月24日~2022年4月14日秦皇岛站铁路的逐日客运量数据，提出了基于季节性差分自回归滑动平均模型(SARIMA)和霍尔特–温特斯模型(Holt-Winters)的铁路客运量组合预测模型。采用方差倒数法，确定各单项模型在组合模型中的权重系数。分别用两个单一模型和组合模型进行15天客运量的短期预测，并采用2022年4月15日~4月29日的实际数据验证预测效果。设置不同的样本梯度，分析样本量对三个模型预测准确度的影响。

Abstract: In the post-epidemic period, the demand for railway passenger transport changed dramatically, and the sudden epidemic interrupted its increasing trend year by year. The establishment of a suitable model to predict passenger volume plays a vital role in the smooth development of railway operation. Based on the daily passenger volume data of Qinhuangdao Station from January 24, 2020 to April 14, 2022, this paper proposes a combined forecast model of railway passenger volume based on seasonal autoregressive integrated moving average (SARIMA) and Holt-Winters model (Holt-Winters). The inverse variance method is used to determine the weight coefficient of each single model in the combination model. Two single models and combination models were used to predict the 15-day passenger volume respectively, and the actual data from April 15 to April 29, 2022 were used to verify the prediction effect. Different sample gradients were set to analyze the influence of sample size on the prediction accuracy of the three models.