A probabilistic approach to detect mixed periodic patterns from moving object data

详细信息    查看全文

• 作者：Jun Li ; Jingjing Wang ; Junfei Zhang ; Qiming Qin ; Tanvi Jindal ; Jiawei Han
• 关键词：Moving object data ; Periodicity detection ; Periodic behavior ; Mixed periodic patterns
• 刊名：GeoInformatica
• 出版年：2016
• 出版时间：October 2016
• 年：2016
• 卷：20
• 期：4
• 页码：715-739
• 全文大小：2,523 KB
• 刊物类别：Earth and Environmental Science
• 刊物主题：Geography
  Geographical Information Systems and Cartography
  Data Structures, Cryptology and Information Theory
  Computer Science, general
  Information Storage and Retrieval
  Multimedia Information Systems
  
• 出版者：Springer Netherlands
• ISSN：1573-7624
• 卷排序：20

文摘

The prevalence of moving object data (MOD) brings new opportunities for behavior related research. Periodic behavior is one of the most important behaviors of moving objects. However, the existing methods of detecting periodicities assume a moving object either does not have any periodic behavior at all or just has a single periodic behavior in one place. Thus they are incapable of dealing with many real world situations whereby a moving object may have multiple periodic behaviors mixed together. Aiming at addressing this problem, this paper proposes a probabilistic periodicity detection method called MPDA. MPDA first identifies high dense regions by the kernel density method, then generates revisit time sequences based on the dense regions, and at last adopts a filter-refine paradigm to detect mixed periodicities. At the filter stage, candidate periods are identified by comparing the observed and reference distribution of revisit time intervals using the chi-square test, and at the refine stage, a periodic degree measure is defined to examine the significance of candidate periods to identify accurate periods existing in MOD. Synthetic datasets with various characteristics and two real world tracking datasets validate the effectiveness of MPDA under various scenarios. MPDA has the potential to play an important role in analyzing complicated behaviors of moving objects.