基于网络货运平台司机驾驶行为的信用评价体系构建  

Construction of Driver Credit Evaluation System Based on Driver Behavior Data from Online Freight Platform

作　　者：杨洋[1] 裴童心 郭丰杰 方羽恬 李艺 YANG Yang;PEI Tongxin;GUO Fengjie;FANG Yutian;LI Yi(School of Management,China University of Mining&Technology(Beijing),Beijing 100083;Beijing Sinoiov InformationTechnology Co.Ltd.,Beijing 100089,China)

机构地区：[1]中国矿业大学(北京)管理学院,北京100083 [2]北京中交兴路信息科技股份有限公司,北京100089

出　　处：《物流技术》2024年第6期119-132,共14页Logistics Technology

基　　金：国家社会科学基金一般项目(22BGL110)。

摘　　要：我国网络货运平台市场规模快速增长,道路货运数字化、规范化、集约化水平不断提升,然而运输过程中部分司机有不良驾驶习惯、服务质量低、运输风险高,导致平台与货主对司机信用评价不高,构建基于网络货运平台司机驾驶行为的信用评价体系有利于降低运输风险,保障司机安全、货物安全和道路交通安全,同时能够提升司机信用水平,提升网络货运平台服务水平。首先,基于司机驾驶行为选择评价指标,通过因子分析提取四类公共因子建立司机信用评价体系;其次,基于SOM-k-means算法对司机驾驶行为进行聚类,从服务质量和运输风险两个角度将司机划分为九种信用等级;最后,通过随机森林模型对不同信用等级进行分值评定,最终确定高价值、中等价值、低价值司机的分数区间,并对不同信用等级的司机提出治理建议。The market size of online freight platforms in China is growing rapidly,and the level of digitali-zation,standardization,and intensification of the road freight transportation industry is constantly rising.However,during transportation,due to the bad habits of some drivers which cause poor service quality and high transportation risks,the platforms and cargo owners have low credit ratings of the drivers,giving promi-nence to the contradiction between the rapid development of the online freight platforms and the imperfect driver credit rating system.Constraining the credit level of platform drivers based on their driving behavior can reduce transportation risks,ensure the safety of the drivers,the cargoes,and the overall road traffic pro-cess,and facilitate the adoption of targeted measures for truck drivers with different credit levels to encourage them to improve service quality and transportation efficiency.First,the driver behavioral profile is calculated based on the driver's waybill trajectory,driving speed,fatigue level and stop data.Common factors including transportation style,transportation risk,transportation efficiency and transportation preference are extracted through factor analysis to establish a driver credit evaluation system.Secondly,the driver's driving credit lev-el is divided based on the SOM-k-means algorithm,and 14 driver types are obtained.From the perspective of service quality and risk level,the 14 types are further summarized into 9 credit types such as the efficient and stable type,mature and reliable type,and efficient and exploratory type,etc.Thirdly,a driver credit scor-ing model is constructed using the random forest algorithm,which converts the characteristic values of actual carriers into corresponding scores,ranking the key factors affecting the driver's credit evaluation in terms of importance,and grouping the driver's credit level into three categories of high value,medium value and low value by calculating the characteristic score.Finally,governance suggestions based on the d

关 键 词：网络货运平台 司机驾驶行为 信用评价体系 SOM-k-means 随机森林 

分 类 号：U492.4[交通运输工程—交通运输规划与管理]