面向多因素影响的混合预测模型  

Mixed prediction model for influence of multiple factors

作　　者：谢宇 王丽清[1] 徐永跃[1] 姚寒冰[1] XIE Yu;WANG Li-qing;XU Yong-yue;YAO Han-bing(School of Information Science and Engineering,Yunnan University,Kunming 650091,China)

机构地区：[1]云南大学信息学院,云南昆明650091

出　　处：《计算机工程与设计》2020年第10期2758-2764,共7页Computer Engineering and Design

基　　金：国家自然科学基金项目(61862064);云南省创新团队基金项目(2018HC019);云南大学服务云南行动计划基金项目(C176240501007)。

摘　　要：为解决在城市道路短时交通流量预测中,由于对复杂外部因素和突发事件的影响考虑不足导致预测误差增大的问题,构建一个多因素混合预测模型。基于ARIMA时序预测模型,通过结合CART模型树引入居民出行习惯、节假日、天气、突发事故等其它外部影响因素,提高预测模型在复杂因素影响下的预测准确度和适应性。实验结果表明,该混合模型在面对节假日和突发事件时,能够得到更准确的预测结果。To solve the prediction error increase problem due to insufficient consideration of influence of complex factors and events in the short-term traffic flow forecast,a mixed prediction model for multi-factors influence was proposed.This model based on the ARIMA time series prediction model,by combining with the CART model tree,the external influence factors were introduced,such as residents’travel habits,holidays,weather,emergencies and other events.The prediction accuracy and adaptability under the influence of multiple factors were improved.Experimental results indicate that the mixed model achieves more accurate prediction results in the face of holidays and emergencies.

关 键 词：差分自回归移动平均模型ARIMA 多因素 预测模型 混合 分类回归树CART 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]