Finding arbitrary shaped clusters with related extents in space and time
The paper deals with density-based clustering of events, i.e. objects positioned in space and time, such as occurrences of earthquakes, forest fires, mobile phone calls, or photos with geographical references. Finding concentrations of events in space and time can help to discover interesting places and time periods. The spatial and temporal properties of event clusters, in particular, their spatial and temporal extents and densities, can be related to each other. Therefore, we suggest a two-step clustering method that considers relations of the spatial and temporal dimension. We demonstrate the work of the method on several examples of real data. 1. Introduction The focus of this work is detecting spatio-temporal clusters (concentrations) of events. By events we mean objects positioned in space (in particular, geographical space) and time. Examples of such objects are occurrences of earthquakes, forest fires, disease cases, mobile phone calls, and photos taken by Flickr or Panoramio users. Concentrations of events in space and time may indicate interesting and/or important phenomena. Places with re-appearing concentrations of events may be interesting or problematic, depending on the nature of the data. Clusters of events may have different shapes in the spacetime continuum. A cluster wide in space but narrow in time means that many events occurred in a large area during a short time interval. Such a cluster in forest fires data may correspond to spreading of forest fires due to weather conditions, in disease data - to a disease outbreak, and in Flickr photos data.