ST-DBSCAN clustering module in SpagoBI for hotspots distribution in Indonesia

Almost every year, Indonesia suffers catastrophic forest fires that has an impact on the shrinking forest land in Indonesia. Hotspot is parameter that can be used as an indicator for forest fires. One of the data mining techniques that can be used to process hotspot data is clustering. Clustering technique is used to obtain interesting patterns so that analysis on the occurrence of hotspots can be done. This research aims to build a clustering module in SpagoBI framework integrated with R engine. The algorithm used in the clustering is ST-DBSCAN with parameters are spatial distance (Eps1), temporal distance (Eps2), and cluster density (MinPts). This research used hotspot data in Indonesia from January to March 2015. The clustering module displays the result and visualizes hotspots clusters in the form of map and chart on HTML page through R software. Analysis of result shows patterns of hotspots' occurrence that appear most frequently are stationary pattern in Bengkalis in January-March 2015.