The Mining of Co-location Patterns with Event-centric Model Approach on Spatial Database

The increasing amount of spatial data provides us with more useful information, either explicitly or implicitly or both. To obtain the implicit information in spatial data, this paper uses a certain technique in Data Mining especially Spatial Data Mining since spatial data is used, it is Co-location Patterns Mining. An example in the application of Co-location pattern mining area is in City management for retrieving Point Of Interest (POI) in proximity neighborhood of instances, such as finding what most instances or objects near hospitals or near a school. This paper evaluates the application of Co-location patterns mining using Event-centric model and Spatial Database using PostGIS to find Point Of Interest (POI). This paper uses DKI Jakarta province data from OpenStreetMap and evaluates 10 major instances on Jakarta. This paper compares two different PostGIS spatial query functions, ST-Dwithin and ST-Distance, to obtain neighborhood relationship in Co-location patterns mining process and to know which function is more efficient or faster. This paper states that ST-Dwithin is faster than ST-Distance and also obtain mosque-school has the biggest Co-location patterns based on Participation Index it has which means that the majority instance near a mosque on Jakarta is school.