摘要
文中对基于决策树优化选择下城市交通出行特征进行研究,通过研究城市交通出行方式,缓解城市交通出行压力。基于决策树算法基本理论,构建决策树模型,选取城市交通出行特征作为分类依据,运用C4.5决策树算法对城市交通出行数据进行分类,根据分类后各个不同特征叶子节点对上层子节点的总占比进行城市交通出行特征优化选择分析,并在“Occam's razor”的基础上,利用重新引入法提出优化方法,解决C4.5决策树算法存在的过度拟合问题,提升城市交通出行方式分析效果。实验结果表明,该方法可有效分析城市交通现有出行特征,指导城市交通规划,依据该方法的分析结果对早高峰线路进行优化后,有效减少了长距离拥堵路段,同时避免了严重阻塞路段的产生。
A study on urban transportation travel characteristics based on decision tree optimization selection is carried out.This study alleviates urban transportation travel pressure by studying urban transportation travel modes.On the basis of the basic theory of the decision tree algorithm,a decision tree model is constructed,and urban traffic travel characteristics are selected as the classification basis.The C4.5 decision tree algorithm is used to classify urban traffic travel data,and the urban traffic travel characteristics are optimized and selected according to the total proportion of each leaf node with different characteristics to the upper sub node after classification.On the basis of″Occam's razor″,the reintroduction method is used to propose optimization methods to solve the overfitting problem of C4.5 decision tree algorithm and improve the analysis effect of urban transportation modes.The experimental results show that the method can effectively analyze the existing travel characteristics of urban traffic and guide urban transportation planning.After optimizing the morning peak line according to the analysis results of this method,it can effectively reduce the long⁃distance congested roads and avoid the generation of serious congested roads.
作者
李文
LI Wen(Hope College of Southwest Jiaotong University,Chengdu 610400,China;Chengdu Transportation+Tourism Big Data Application Technology Research Base,Chengdu 610400,China)
出处
《现代电子技术》
北大核心
2024年第5期182-186,共5页
Modern Electronics Technique
关键词
城市交通
出行特征
决策树
优化选择
特征分类
C4.5决策树算法
奥卡姆剃刀理论
过度拟合
urban transportation
travel characteristic
decision tree
optimization selection
characteristic classification
C4.5 decision tree algorithm
Occam's razor theory
overfitting