摘要
多源时空轨迹数据隐含丰富的城市出行信息,通过对其进行挖掘、处理和分析,可以找到个体与群体之间的交互关系。针对轨迹数据挖掘研究范围单一,缺少多空间尺度研究的问题,提出一种融合多空间尺度特征的出行轨迹数据挖掘分析方法。以广东为例,结合社交媒体腾讯用户密度(Tencent user density,TUD)数据集,通过具有噪声的基于密度的聚类方法(density-based spatial clustering of applications with noise,DBSCAN)聚类算法与局部密度峰值计算法提取时空相似性轨迹区域,进而簇类分成一系列热点区域,获得不同时间粒度、不同空间尺度下的出行轨迹规律特征。这能够实现在不同空间尺度融合下展示同一地区的热点区域,进一步探讨出行轨迹的规律变化。可见所提出的方法为利用时空大数据进行城市空间结构研究提供科学参考。
A wealth of urban travel information is implied in the multi-temporal trajectory data,and the relationships between individuals and groups can be found through mining,processing,and analyzing.A travel trajectory data mining and analysis method based on multi-spatial scale features was proposed to solve the problems of the single scope of trajectory data mining study and the lack of multi-spatial scale research.Taking the Tencent user density(TUD)dataset in Guangdong Province as an example,the density-based spatial clustering of applications with noise(DBSCAN)clustering algorithm and local density peak calculation method were used to extract spatio-temporal similarity trajectory areas,then the traveling trajectory pattern features of different temporal granularity and spatial scales were obtained by clustering into a series of hotspot areas.These results could display the hotspots in the same area of different spatial scale fusion,and the regular changes of traveling trajectories can be further explored.It is concluded that the proposed method can provide a scientific reference to urban spatial structure research by spatio-temporal big data.
作者
陆妍玲
黄娅琦
王杰
黄露
赵毅
李景文
LU Yan-ling;HUANG Ya-qi;WANG Jie;HUANG Lu;ZHAO Yi;LI Jing-wen(College of Geomatics and Geoinformation,Guilin University of Technology,Guilin 541004,China;Guangxi Key Laboratory of Ecological Spatio-temporal Big Data Perception,Guilin 541004,China)
出处
《科学技术与工程》
北大核心
2023年第20期8530-8539,共10页
Science Technology and Engineering
基金
国家自然科学基金(41961063)。
关键词
多空间尺度
具有噪声的基于密度的聚类方法(DBSCAN)算法
局部密度峰值
热点区域
时空分析
multiple spatial scales
density-based spatial clustering of applications with noise(DBSCAN)algorithm
local density peak extraction
hot spots
analysis of spatio-temporal
