基于GPS数据的用户轨迹挖掘与预测研究

发布时间：2018-04-26 22:38

本文选题：轨迹挖掘 + 行为预测　；参考：《北京邮电大学》2017年硕士论文

【摘要】：近年来,随着移动互联网技术的迅猛发展和智能移动终端的普及,各种丰富多彩的智能移动应用和服务正在呈现爆炸性增长的态势,逐渐成为人们日常生活的一部分。与此同时,这些移动应用与业务产生了海量的与位置相关的交互数据与商务数据等,如何对这些海量数据进行数据挖掘,沙里淘金,提取出有价值的信息,已经成为当下的研究热点。基于GPS数据的用户轨迹挖掘与用户行为预测研究,具有很多的应用场景。针对单个用户来说,通过对用户历史轨迹数据的数据挖掘、机器学习,可以总结预测用户兴趣、职业、行为习惯等信息,然后为用户提供个性化推送式的服务,如智能旅游向导、位置提醒服务、音乐推荐、精准广告推送等等;针对用户人群来说,通过对人群数据的建模学习,可以获取地区的热门景点、热门饭店、热门活动以及不同用户之间的人脉关系等,这对智能城市规划、智能交通、热门活动监控等具有很大应用价值。现有的位置服务大多直接使用用户提供的数据,缺乏对用户轨迹数据的深层次挖掘研究,无法获取单个用户的行为习惯与用户人群的流动倾向,提供有针对性的个性化服务,本文主要是基于用户的移动轨迹数据,结合地理信息数据,深入地研究了如何聚类提取单个用户或用户群体的兴趣点,如何提取、矫正用户移动轨迹,如何基于位置对用户的行为习惯进行建模预测等内容,借鉴国内外先进的研究成果,提出了基于密度聚类的兴趣点聚类算法、基于流式的兴趣点聚类算法,基于路网匹配的路径提取与矫正算法以及基于BP神经网络模型的用户位置行为预测算法。通过具体的实验分析,本论文提出的算法方案具有很好的可靠性、准确性与创新性。
[Abstract]:In recent years, with the rapid development of mobile Internet technology and the popularity of intelligent mobile terminals, a variety of colorful intelligent mobile applications and services are showing an explosive growth trend, and gradually become a part of people's daily life. At the same time, these mobile applications and business have produced massive interactive data and business data related to location, etc. Has become the current research hotspot. The research of user trajectory mining and user behavior prediction based on GPS data has many application scenarios. For a single user, through the data mining and machine learning of the user's historical track data, we can sum up and predict the user's interest, occupation, behavior habits and other information, and then provide personalized push service for the user. Such as intelligent travel guide, location remind service, music recommendation, accurate advertising push and so on; for users, through the modeling of crowd data learning, you can get the hot spots in the area, hot hotels, The popular activities and the relationships between different users have great application value for intelligent city planning, intelligent transportation, monitoring of popular activities and so on. Most of the existing location services directly use the data provided by users, lack of deep mining research on user trajectory data, unable to obtain individual users' behavior habits and user mobility tendency, and provide targeted personalized services. This paper is mainly based on the user's mobile trajectory data, combined with geographic information data, in-depth research on how to cluster to extract individual users or user groups of interest points, how to extract, correct the user mobile trajectory, How to model and predict the behavior habits of users based on location, and draw lessons from the advanced research results at home and abroad, this paper puts forward the clustering algorithm of points of interest based on density clustering, and the clustering algorithm of points of interest based on flow. The path extraction and correction algorithm based on road network matching and the prediction algorithm of user's position behavior based on BP neural network model. Through the experimental analysis, the algorithm proposed in this paper has good reliability, accuracy and innovation.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P228.4;TP311.13

【相似文献】