Segmentation of Thai Provinces Using Fire Incident Time Series Data

Fire incidents can be categorized into predictable and unpredictable events. Predictable events include agricultural burning, while unpredictable events refer to wildfires. These incidents can be detected by satellite systems using the so-called hotspot fire. Such events cause significant environmental damage and contribute to air pollution, particularly the PM 2.5 problem we are currently facing. In this work, we apply hierarchical clustering using a recently introduced dissimilarity measure, called RDPC, to segment Thailand's provinces using fire incident time series data. We end up with eleven distinct groups of provinces with different characteristics. Finally, we suggest some preventive plans specifically for each cluster.