摘要
为提升网约车行业管理精度及运力评估准确性,本文基于苏州市2024年9月订单数据,采用K-means聚类算法对网约车司机的运营行为进行分类识别;以司机出车频率、工作时长及营收水平等六项指标为基础,构建司机特征数据集,并通过轮廓系数法确定聚类数目。聚类结果显示,司机可划分为兼职与全职两类,其中:兼职司机出车时间短、营收水平低,接单主要集中于工业区等生活圈范围;全职司机运营时间长、服务能力更强,接单集中在人流密集区域如交通枢纽、商业区等;在订单数量与实际运力贡献上,二者存在数量与效率的不对称性。研究结果表明,仅以司机总量衡量运力并不准确,建议行业监管引入“有效供给”视角,推动差异化管理策略落地,实现精准调控和资源优化配置。
To enhance the precision of mobility resource management in the online ride-hailing industry,this study applies K-means clustering to classify the operational behaviors of drivers in Suzhou using platform order data from September 2024.Six core indicators––weekly working days,average daily working hours,and average daily revenue are selected to construct the driver feature profiles.The optimal number of clusters is determined using the silhouette coefficient.Results show that drivers can be effectively categorized into two groups:part-time and full-time.Part-time drivers exhibit low frequency and limited geographic service areas,mostly operating around industrial zones.Full-time drivers,in contrast,provide more consistent and extensive service,especially in high-demand areas such as transportation hubs and commercial districts.Notably,full-time drivers contribute significantly more to overall trip volume.The findings indicate that total driver count does not accurately reflect actual transportation supply.It is recommended that policymakers incorporate the concept of"effective supply"and adopt differentiated regulation strategies to better align resources with demand.
作者
吾晨晨
庄楚天
WU Chenchen;ZHUANG Chutian(Suzhou Planning&Design Research Institute Co.,Ltd,Suzhou 215000,China;CCDI(Suzhou)Survey&Design Consultant CO.,Ltd.,Suzhou 215123,China)
出处
《交通与运输》
2025年第4期86-91,共6页
Traffic & Transportation
关键词
网约车司机
运营特征
聚类分析
兼职与全职
接单行为
运力管理
ride-hailing drivers
operational characteristics
K-means clustering
part-time and full-time drivers
order-taking behavior
mobility supply management
