Prophet-X-12-ARIMA组合模型及交通运输量预测

Prophet-X-12-ARIMA Combined Model and Transport Volume Prediction

由于受到经济社会因素的影响,交通运输量数据具有趋势多变及季节性明显的复杂特征。现有的预测模型,如X-12-ARIMA模型、ARIMA模型和Prophet模型等的预测准确性有待改进。文章构建Prophet-X-12-ARIMA组合模型,综合了Prophet模型灵活拟合趋势成分的优势以及X-12-ARIMA模型能准确分解出季节成分的优点。采用该模型预测某城市的七种交通运输量序列,结果显示Prophet-X-12-ARIMA组合模型的半年度和年度预测效果明显优于Prophet模型、X-12-ARIMA模型及ARIMA模型。进一步研究发现,当原始序列趋势变化剧烈时,Prophet-X-12-ARIMA组合模型的预测效果更优。

Due to the influence of economic and social factors, the traffic volume data has the complex characteristics of changing trend and obvious seasonality.The prediction accuracy of existing prediction models, such as X-12-ARIMA model, ARIMA model and Prophet model, needs to be improved. To this end, this paper constructs the Prophet-X-12-ARIMA combined model, which integrates the advantages of the Prophet model for flexible fitting of trend components and X-12-ARIMA model for accurate decomposition of seasonal components. The results of using the model to predict seven traffic sequences of a city show that the semi-annual and annual forecasting effects of the Prophet-X-12-ARIMA combined model are significantly better than the Prophet model, the X-12-ARIMA model and the ARIMA model. Further study finds that when the trend of the original sequence changes drastically, the prediction effect of the Prophet-X-12-ARIMA combined model is better.