On Spatio-temporal Events Clustering Methods

Clustering analysis of spatio-temporal event is a forefront research in the spatio-temporal data mining domain.It has important applications in disease early warning and controlling,climate change,earthquake prediction and the analysis of crimes.Firstly,the research actuality and new progresses in spatio-temporal event clustering algorithm in recent years are investigated and summarized.Secondly,the analysis and comparison of four representative spatio-temporal event clustering methods(i.e.Space-time permutation scan statistic,STDBSCAN,WNN,STSNN) have been made from the views of clustering quality and user operation.Moreover,clustering conditions of the same method for different types of data sets as well as different methods for the same datasets are analyzed.Finally,the advantages,disadvantages and applicability of these four clustering methods are summarized after several experiments and comparative analyses,and a number of issues for further research are highlighted.