Geo-spatial analytics using the dynamic ST-SNN Approach

Abstract

Spatio-temporal clustering is a subfield of data mining that is increasingly gaining more scientific attention due to the advances of location-based devices that register position, time and, in some cases, other attributes. Spatio-temporal clustering intends to group objects based in their spatial and temporal similarity helping to discover interesting spatio-temporal patterns and correlations in large data sets. One of the main challenges of this area is the ability to integrate spatial, temporal and other numerical or classification information in a general-purpose approach as well as the capability to integrate, in the previously obtained clusters, newly available data. This paper presents the Dynamic ST-SNN approach in which the user has the possibility to simultaneously analyse several dimensions and incrementally add new-collected data to the existing clusters providing updated clusters.