基于大规模GPS轨迹数据的出租车服务策略研究

发布时间：2019-02-11 13:43

【摘要】：目前,在城市交通领域,GPS设备已被广泛应用。例如,几乎每辆出租车均安装了GPS终端设备,这些设备会将出租车的行驶位置和运营状态等数据传输给交通管理部门用于实时监控。事实上,在出租车GPS数据中隐藏着出租车司机的群体智能信息如服务策略等,充分利用这些信息,一方面能够引导出租车司机改善运营方式并增加收入,另一方面也能够帮助管理部门提高出租车系统运营效率。然而,目前对出租车GPS数据的分析手段还相对初级;同时,出租车GPS数据规模庞大,例如,西安市一年的出租车GPS数据规模接近TB级。实践证明,传统的信息处理平台已经无法对上述规模的GPS数据进行有效的分析。针对上述问题,本文基于大数据平台Hadoop对出租车GPS数据进行出租车服务策略的挖掘和分析。Hadoop是目前流行的一种大数据平台,采用并行计算结构能够完成TB级以上规模的数据分析工作,同时由于开源、可扩展、低成本、易于编程等优点,已经成为一个大数据处理领域的事实标准。出租车服务策略是针对司机服务过程建立的数学模型,主要指乘客搜索策略、乘客输送策略、服务区域偏好等。在对GPS数据进行数据预处理之后,首先提取出租车白班运营轨迹,并根据GPS轨迹对出租车司机的收入进行量化;根据收入排名取出收入较高和一般的司机群体作为样本司机,提取并统计样本司机的服务策略及使用情况,最后分析服务策略与收入的关系。在测试实例中,本文对样本司机不同时隙在乘客搜索策略、乘客输送策略、区域偏好上存在的异同点进行了分析和比较。
[Abstract]:At present, in the field of urban transportation, GPS equipment has been widely used. For example, almost every taxi is equipped with GPS terminal equipment that transmits data such as taxi location and operation status to traffic management for real-time monitoring. In fact, the taxi drivers' intelligence information, such as service strategy, is hidden in the taxi GPS data. On the one hand, it can guide taxi drivers to improve their operation mode and increase their income. On the other hand, it can also help the management to improve the efficiency of taxi system operation. However, the current analysis of taxi GPS data is relatively preliminary; at the same time, taxi GPS data scale is huge, for example, Xi'an taxi GPS data scale is close to TB level. It has been proved that the traditional information processing platform can not effectively analyze the GPS data of the above scale. In order to solve the above problems, this paper uses big data platform Hadoop to mine and analyze taxi service strategy of GPS data. Hadoop is a popular big data platform. Using parallel computing structure can complete the data analysis work of TB scale and above. At the same time, because of the advantages of open source, expandability, low cost and easy programming, it has become a de facto standard in big data's processing field. Taxi service strategy is a mathematical model for driver service process, which mainly refers to passenger search strategy, passenger transportation strategy, service area preference and so on. After the data preprocessing of GPS data, the operation track of taxi day shift is extracted firstly, and the income of taxi driver is quantified according to the GPS track. According to the rank of income, the higher income and general group of drivers are taken as sample drivers, and the service strategy and usage of the sample drivers are collected and counted. Finally, the relationship between service strategy and income is analyzed. In the test example, the similarities and differences in passenger search strategy, passenger transport strategy and regional preference between different time slots of sample drivers are analyzed and compared.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495;TP311.13

【参考文献】