摘要
通过分析城市轨道交通日均客流及相关影响因素的变化特征,以多种因素作为数据集特征,采用梯度提升法和随机森林的混合模型对日均客流量进行预测。以北京地铁客流数据作为研究对象,对模型进行了试验。试验结果表明,基于梯度提升和随机森林的混合模型相较于常规ARIMA模型和随机森林模型具有更好的适应性,在常态和特殊情况下均能取得可接受的预测效果。
By analyzing the daily rail transit passenger flow and the variation characteristics of related influencing factors,various factors are collected as the data set features,and a gradient boosting regression method is employed together with the random forest model to forecast the daily passenger flow.Then,the model is tested based on the data of Beijing metro,the result shows that the hybrid model based on gradient boosting and random forest has better performance than the conventional ARIMA and random forest model,the prediction result of the former is acceptable in both normal and special conditions.
作者
丁聪
倪少权
吕红霞
DING Cong;NI Shaoquan;LYU Hongxia(School of Electronics Engineering and Computer Science,Southwest Jiaotong University,611756,Chengdu,China)
出处
《城市轨道交通研究》
北大核心
2018年第9期60-63,共4页
Urban Mass Transit
关键词
城市轨道交通
客流量预测
梯度提升
urban rail transit
passenger flow forecast
gradient boosting