Spatio-temporal graph clustering algorithm based on attribute and structural similarity

The rapid growth of data in Spatio-temporal datasets collected from several domains such as climate science, Health Science, Sensor networks and telecommunication systems has created a need for Spatio-Temporal clustering methods to extract and analyse the dynamic clusters. Detecting dynamic clusters based on spatial dependence of objects with heterogeneous properties over space and time is a challenging task in clustering spatio temporal datasets. Spatio-Temporal Graph Clustering Algorithm is proposed for detecting communities in Health dataset. The people infected by H1N1 flu virus (Swine Flu) in India is considered as the study area and dataset for the analysis. Mobility of an infected individual in epidemiology plays a key role in modeling of disease spread. The climatic condition of the location is considered as the attributes and mobility rate between the locations are treated as the edge weight. The change in transition of each spatial node is analysed statistically using Local Spatial Autocorrelation test to find the dependency of spatial node dynamically for each time period. Four communities are detected and ranked based on the structural and attribute similarity that is achieved using neighbourhood of the spatial node and similarity score of each node. Performance of the algorithm is validated with the existing algorithm. This study helps the administrative officials to take rapid preventive measures like vaccination strategies to control the spread of disease.