基于生存分析的通勤行为出发时间研究

发布时间：2019-01-03 16:38

【摘要】：通勤出行不仅在城市居民一日出行行为中所占比例较高,同时也是居民日常出行中最基础、最重要的行为之一。特别是在通勤出行集中的高峰期和区域,很容易诱发形成早、晚高峰通勤出行的交通拥堵。通勤行为出发时间是反映通勤交通自身规律及时空分布的重要参数,因此有必要对其进行深入研究。本文以“通勤行为的出发时间”为切入点,从个体角度来考虑通勤行为出发时间的影响因素。本文通过问卷调查获取通勤者的通勤出行行为数据,基于生存分析理论,提取了影响通勤者出发时间的诸多因素,并通过变量筛选、模型假定条件筛选等环节,建立通勤行为出发时间的Cox风险模型,根据模型的回归结果得到各变量对通勤行为出发时间的具体影响作用,并用该模型预测了个体出发时间的选择情况。本文的主要研究内容具体如下:(1)通勤行为出发时间的总体分布规律。基于调查问卷数据,运用生存分析的非参数法对通勤行为出发时间的生存函数和风险函数作估计,探索通勤出发时间的总体分布规律。(2)通勤行为出发时间的Cox风险模型及其影响因素分析。讨论Cox比例风险模型用于通勤行为出发时间研究的适用性,建立通勤行为出发时间的Cox风险模型,对可能存在的影响因素进行筛选后,最终确定性别、年龄、职业、是否送学生就学、出行距离、出行方式、公共交通服务质量及个体机动化交通服务质量共8个因素均会对通勤出发时间产生较显著的影响。(3)基于通勤者的家庭特性、个人特性、交通特性以及个体感知的交通服务质量,应用Cox风险模型对通勤行为出发时间进行预测,结果表明该模型的预测性较好,能够比较准确的预测通勤行为的出发时间。本文的研究充实了通勤行为出发时间个体层面的现有研究,能够为交通诱导措施、交通管理控制措施的制定提供可借鉴的理论依据。
[Abstract]:Commuting is not only a high proportion of daily travel behavior of urban residents, but also one of the most basic and important behaviors. Especially in the rush hour and area of commuting concentration, traffic congestion of early and late commuting is easy to be induced. The departure time of commuting behavior is an important parameter to reflect the time and space distribution of commuting traffic, so it is necessary to study it deeply. This paper takes "departure time of commuting behavior" as the starting point, and considers the influence factors of commute behavior departure time from the angle of individual. In this paper, the commuter travel behavior data are obtained by questionnaire survey. Based on the survival analysis theory, many factors affecting the commuters' departure time are extracted, and some links, such as variable selection, model assumption condition selection, etc. The Cox risk model of commuting behavior departure time is established. According to the regression results of the model, the specific effect of each variable on the commuting behavior departure time is obtained, and the choice of individual departure time is predicted by the model. The main contents of this paper are as follows: (1) the general distribution of commuting behavior departure time. Based on the questionnaire data, the survival function and risk function of commuting behavior were estimated by the nonparametric method of survival analysis. To explore the general distribution of commuting departure time. (2) the Cox risk model of commuting behavior departure time and the analysis of its influencing factors. The applicability of Cox proportional risk model for the study of commuting behavior departure time was discussed. The Cox risk model of commuting behavior departure time was established. After the possible factors were screened, the sex, age and occupation were determined. Whether to send students to school, travel distance, travel mode, public transportation service quality and individual motorized traffic service quality all have significant effects on commuting departure time. (3) based on the family characteristics of commuters, The Cox risk model is used to predict the departure time of commuting behavior. The results show that the model is more predictable and can accurately predict the departure time of commuting behavior. The research in this paper enriches the existing research on commuting behavior departure time at the individual level and can provide the theoretical basis for the formulation of traffic guidance measures and traffic management control measures.
【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】：U491

【参考文献】