基于地理加权回归的城市出租车客流影响因素分析及建模

发布时间：2018-03-10 02:14

本文选题：出租车客流　切入点：空间异质性　出处：《西南交通大学》2017年硕士论文　论文类型：学位论文

【摘要】：出租车运输行业凭借独有的优势,一直以来在城市交通旅客运输市场占据一席之地,出租车已经成为城市交通系统中重要组成之一。可靠的出租车客流需求预测可以帮助交通管理部门和运输服务企业对车租车市场的合理调控。一方面,可提高交通管理部门和运输服务企业对资源的合理分配,提高盈利;另一方面,居民出行选择将更加灵活自由。然而,随着城市的多点多级发展以及区域发展的差异,城市出租车客流分布的规律在空间局部的表现也不相同。本文利用纽约市出租车数据资料,通过分析城市出租车客流与各种空间上明确的社会人口和内置环境变量之间的关系,对其空间变化进行研究。运用地理加权回归模型(GWR)建立出租车客流的空间异质性模型,并且将估计参数的空间变化可视化。主要研究内容如下:1)根据出租车运输特点对城市出租车客流需求进行相关基础理论研究;2)重点研究地理加权回归模型的结构,回归模型参数的估计方法;3)使用数据库和GIS软件对数据筛选、计算;4)使用处理得来的数据建立了基于地理加权回归的出租车客流影响因素分析模型;5)将模型估计参数的空间变化可视化,深入分析各影响因素对局部区域出租车客流的影响。研究结果表明,在处理分析空间变化的问题上,地理加权回归模型的模型拟合度和变量解释准确度都优于普通多元线性回归模型;城市的空间结构对出租车客流分布的影响显著,并且参数估计值存在较为明显的空间异质性。这些结果为预测出租车客流作为空间变量的函数提供了有价值的见解,这可以对出租车定价,出租车行业监管和城市规划产生重要意义。
[Abstract]:By virtue of its unique advantages, the taxi transportation industry has always occupied a place in the urban transportation and passenger transportation market. Taxi has become one of the important components of the urban transportation system. Reliable taxi passenger demand prediction can help traffic management departments and transport service enterprises to regulate the car rental market reasonably. On the one hand, On the other hand, residents will be more flexible and free in their travel choices. However, with the multi-point and multi-level development of the city and the difference of regional development, The distribution law of urban taxi passenger flow is also different in spatial part. This paper analyzes the relationship between urban taxi passenger flow and various spatial clear social population and built-in environmental variables by using the taxi data of New York City. The spatial variation of taxi passenger flow is studied. The spatial heterogeneity model of taxi passenger flow is established by using geographical weighted regression model (GWR). And visualize the spatial variation of the estimated parameters. The main research contents are as follows: (1) based on the characteristics of taxi transportation, the basic theory of urban taxi passenger demand is studied. (2) the structure of the geo-weighted regression model is studied. Estimation of regression model parameters using database and GIS software to filter the data, Calculation 4) using the processed data to establish the analysis model of the influencing factors of taxi passenger flow based on geographical weighted regression (5) to visualize the spatial variation of the estimated parameters of the model. The effects of various factors on the taxi passenger flow in local areas are analyzed in depth. The results show that, in dealing with the problem of spatial change, The model fitting degree and variable interpretation accuracy of the geo-weighted regression model are better than those of the ordinary multivariate linear regression model, and the spatial structure of the city has a significant effect on the taxi passenger flow distribution. These results provide valuable insights for predicting taxi passenger flow as a function of spatial variables, which can be used to price taxis. Taxi industry regulation and urban planning have important significance.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U492.434

【参考文献】