我国铁路客运量短期预测模型修正及比较

Modification and Comparison of Short-term Forecasting Models for Railway Passenger Volume in China

文章针对我国铁路客运量进行短期预测,首先分析现有常见预测模型的优点与不足,然后试图通过构造新的组合或修正模型,从而实现提高预测精度。在构建年度数据组合模型时,发现以偏最小二乘回归、主成分回归和岭回归为基础进行组合时,预测精度达到了最优;在构建季度数据模型时,首先通过修正的时间序列分解法与季节周期回归模型显著地提高了预测精度,然后以这两个模型为基础构造组合模型,预测精度进一步得到提高。

This paper forecasts the passenger traffic volume of China’s railway in the short term. Firstly, the paper analyzes the advantages and disadvantages of the existing common prediction models, and then tries to improve the prediction accuracy by constructing new combination or modification models. When the annual data combination model is being constructed, it is found that the prediction accuracy is optimal when partial least square regression, principal component regression and ridge regression are combined. In the construction of quarterly data model, the prediction accuracy is significantly improved by the modified time series decomposition method and seasonal cycle regression model. Then the combined model is constructed based on these two models and the prediction accuracy is further improved.