基于电动自行车数据的用户行为分析及可视化研究

发布时间：2018-06-23 11:49

本文选题：电动自行车数据 + 居住地挖掘　；参考：《浙江大学》2017年硕士论文

【摘要】：近年来,电动自行车行业迅速发展,很大程度上改善了居民出行条件,提高了出行效率。但与之相关的交通问题、社会治安问题也日益凸显,例如扰乱交通秩序、引发交通事故、车辆频繁被盗等。为了解决电动自行车发展带来的一系列问题,管理部门需要全面深入地了解该群体的行为特征,以加强对电动自行车的管理,使其更好地服务于社会经济的发展。当前,我国部分城市安装了电动自行车智能防盗追踪系统,采集了大量的电动自行车实时位置信息,使我们可以从海量数据出发,利用大数据相关技术,挖掘其中蕴含的有价值信息,为电动自行车管理提供指导建议。本文基于我国某城市的电动自行车数据开展了系统性研究工作,对电动自行车用户的行为特征进行深入分析,并构建了功能丰富的可视化交互系统。具体贡献如下:本文基于电动自行车数据提取停留地点和移动轨迹,建立并实现了基于分层模型的用户居住地自动挖掘算法。为分析用户对不同停留地点的访问规律,采用停留地点偏好矩阵建模用户的日程访问行为,并提出一种度量用户日程访问行为相似性的方法,在此基础上建立Kmeans用户聚类模型,采用DB Index确定最优聚类类别。实验表明模型可有效地区分用户行为规律,帮助人们了解其生活习惯与职业背景分布。为利用电动自行车移动轨迹估算电动自行车骑行速度,本文针对现有道路匹配算法不能解决采样频率低、空间相关性弱的电动自行车数据的道路匹配问题,设计了一种基于路径约束的道路匹配算法。该算法通过多个数据间的空间联系确定行驶路径,并综合利用kd树、A*搜索算法和动态规划方法,在提高匹配准确性的同时保证了较高的计算效率。在数据分析的基础上,本文还构建了Web端的可视化系统,方便人们直观地分析和理解电动自行车数据。该系统包括电动自行车监控、移动性分析和用户行为分析等多种功能,为本文研究工作的实际应用提供了可能。
[Abstract]:In recent years, the rapid development of electric bicycle industry has greatly improved the travel conditions of residents and increased travel efficiency. However, the related traffic problems and social security problems have become increasingly prominent, such as disturbing traffic order, causing traffic accidents, frequent theft of vehicles and so on. In order to solve a series of problems brought about by the development of electric bicycle, the management department should fully and deeply understand the behavior characteristics of this group, in order to strengthen the management of electric bicycle and make it serve the development of social economy better. At present, some cities of our country have installed the intelligent anti-theft tracking system of electric bicycle, and collected a large amount of real-time location information of electric bicycle, so that we can start from the massive data and make use of the related technology of big data. Excavate the valuable information contained therein, provide the guidance suggestion for the electric bicycle management. Based on the data of electric bicycle in a certain city of our country, this paper carries out systematic research work, deeply analyzes the behavior characteristics of electric bicycle users, and constructs a visual interactive system with rich functions. The main contributions are as follows: based on the data of electric bicycle, an automatic mining algorithm of user residence based on hierarchical model is established and implemented in this paper. In order to analyze the rules of users' access to different places of stay, this paper presents a method to measure the similarity of user's visit behavior of schedule, and then establishes a Kmeans clustering model. DB Index is used to determine the optimal clustering category. Experiments show that the model can effectively distinguish user behavior and help people understand their living habits and professional background distribution. In order to estimate the cycling speed of electric bicycle by using the moving track of electric bicycle, the existing road matching algorithm can not solve the road matching problem of the data of electric bicycle with low sampling frequency and weak spatial correlation. A path matching algorithm based on path constraint is designed. The algorithm determines the driving path by the spatial connection of multiple data, and synthetically uses the KD tree search algorithm and dynamic programming method to improve the accuracy of the matching and ensure a high computational efficiency. On the basis of data analysis, this paper also constructs a visual system of Web side, which is convenient for people to analyze and understand the data of electric bicycle intuitively. The system includes many functions, such as electric bicycle monitoring, mobility analysis and user behavior analysis.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U491.225;TP311.13

【参考文献】