基于出租车运营数据的载客区域聚类及热点特征分析

Clustering Of Passenger Picked-Up Areas and Analysis of Hot Spots Based on Taxi Business Data

摘要: 为优化出租车空间资源调度，针对路网密集、数据密度差异较小条件下的城市出租车载客区域聚类问题，本文提出了一种采用OPTICS算法的聚类模型。通过实验与传统DBSCAN算法下的聚类结果进行对比，发现OPTICS算法更能有效地避免参数设置对实验结果的影响，解决传统DBSCAN算法在此类应用中聚类划分模糊的问题，达到出租车资源调度精细化的效果，对提升出租车载客率、降低空载时间具有现实指导作用。

Abstract: To optimize the taxi space resource dispatch, a clustering model using OPTICS algorithm is pro-posed for the clustering of taxi passenger carrying area under the condition of intensive road and small difference in data density. By comparing with the results of the traditional DBSCAN algorithm in experiment, it is found that the OPTICS algorithm can effectively eliminate the interference of parameter setting on the experimental results. It helps to the problem of traditional algorithm DBSCAN applied to such situation, and enhances the efficiency of taxi dispatch. This research has practical guiding effect on improving the load factor of taxi and reducing the time of idle load.