基于图数据库的公交出行行为分析

Research on transit trip behavior based on graph database

文章基于图数据库Neo4j构建了公交出行行为分析场景建模,对深圳市约1000条公交线路,一天工作日的公交刷卡数据进行建模分析,建模规模达到500万节点,1000万条边。研究分析了同乘人员、站点最大客流提取、关联查询识别、站点群公交出行量识别等公交出行行为。在不同的公交出行行为场景分析中,对比分析了ORACLE数据库与Neo4j的查询性能,多维关联查询中Neo4j性能远高于ORACLE数据库,表明基于图数据库可方便高效实现复杂数据的关联挖掘分析。

Based on the graph database Neo4j,this paper constructs a traffic behavior analysis scenario modeling of the bus and analyzes the bus credit card data of about 1 000 bus lines and one-day working days in Shenzhen. The modeling scale reaches 5 million nodes and 10 million sides. The study analyzes the bus travel behaviors such as passengers,maximum passenger flow extraction,associated query identification,and bus group travel identification. In the analysis of different bus travel behavior scenarios,the query performance of ORACLE database and Neo4j is compared and analyzed. The performance of Neo4j in multi-dimensional relational query is much higher than that of ORACLE database,which indicates that the graph database can conveniently an d efficiently realize the association mining analysis of complex data.