Trajectory Pattern Mining Using Sequential Pattern Mining and K-Means for Predicting Future Location

International Conference on Computing and Applied Informatics 2016

物理学;计算机科学

Kautsar, G.^1 ; Akbar, S.^1

Informatics, School of Electrical Engineering and Informatics, Institut Teknologi Bandung, Ganesha Street No. 10, Bandung, Indonesia^1

关键词: ITS data;    Movement history;    Moving objects;    Sequential-pattern mining;    Spatial attribute;    Temporal attributes;    Time duration;    Trajectory pattern;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/801/1/012017/pdf
DOI  :  10.1088/1742-6596/801/1/012017

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

Sequential pattern mining is a method used to find patterns while concerning the sequence of an item set. Sequential pattern mining can be used to find trajectory patterns in moving object data. To implement it in the real life, the spatial attribute of the data needs to be generalized/grouped. In this paper, K-Means is used to group the spatial attribute. In order to group the spatial attribute, the temporal attribute is also considered to see how the patterns are related to time. The resulting trajectory patterns are then used to visualize the habit of the moving object. Therefore, trajectory patterns are used as the reference in this paper to predict the future location of the object. Predicting the future location of the object is performed using the movement history of the object. Result of this research is trajectory pattern which repeat at certain time duration according to its data characteristics.