Penerapan Dynamic Density Based Clustering pada Data Kebakaran Hutan
Abstract
Land and forest fire has become prominent issues in Indonesia. Possibilities of the occurrence of forest fire in Indonesia has increased from year to year. This makes early prevention very important for forest fire investigation. One of the efforts concerning the forest fire prevention is by knowing the distribution of hotspot clustering which have high potential for the occurrence of forest fire. This research has classified the hotspot data using Dynamic Density Based Clustering (DDBC) algorithm. The use of DDBC technique is capable of handling spatiotemporal aspects simultaneously by storing the position of each point. The storage of each point’s position is estimated using the strength of its relationship to other points that appear every year. The neighborhood concept of DDBC algorithm is a modified version of the neighborhood concept of the Density Based Spatial Clustering (DBSCAN) called Relationship Strength Threshold (RST). Cluster detection is performed on the points that fulfill the RST neighborhood value, so that only the point which was considered as a strong relationship will be grouped. The result of the clustering obtained through DDBC technique is the grouping of areas with high potential for forest fire occurrence. Visualization of the clustering results is presented based on a map that describe the distribution of hotspot so that the authorities can determine the prioritized areas for early forest fire prevention.
Collections
- UT - Computer Science [2255]