Sequential Patterns for hotspot occurrences based weather datausing Clospan algorithm

Abstract

Weather is one of some contributing factors causing forest fires. A hotspot is an indicator of forest fires. Weather and hotspots data can generate sequential patterns of hotspot occurrences based on weather data. The sequential pattern can be used in making right decisions or policies to prevent forest fires. This work applied the Closed Sequential Pattern Mining (Clospan) algorithm that is available in the Sequential Pattern Mining Framework (SPMF)program to generate sequential patterns on hotspots data. The data used are hotspots, precipitation and temperature that are grouped by year of events starting from the year 2001 to 2010. The sequential patterns were discovered with minimum supports from1% to 20%. The results show that the sequential patterns generated from hotspots and precipitation data indicate the first hotspot occurrence in a location with precipitation 0.03 inch per 6 hours followed by precipitation 0.20 inches per 6 hours at different times. Sequential patterns of hotspot and temperature data indicate the first hotspot occurrence in a location with temperature 28.33 °C followed by temperature 28.89 °C and temperature 29.44 °C at different times. Areas where most hotspots commonly found are those with precipitation 0.03 inch per 6 hours and temperature 29.44 °C.