Association rules mining on forest fires data using FP-Growth and ECLAT algorithm

Forest fires and land are a serious problem that must be solved by the Indonesian government including Riau Province. One of forest fires prevention effort is discovering relationship patterns of hotspot occurrences as fire indicators with characteristics of geographic objects where the hotspots occur. The objective of this research is to apply the multidimensional association rule mining method with Frequent Pattern Growth algorithm (FP-Growth) and Equivalence Class Transformation algorithm (ECLAT) to determine association patterns between hotspot occurrences and its supporting factors. The factors that influence hotspot occurrences were discovered on minimum support of 30% and minimum confidence of 80% with hotspot occurrence as the target attribute. The result of this research shows that strong relationships between hotspot occurrences and its influence factor were found with the the highest support of 44.49%, confidence of 100%, and lift of 1.02, where hotspot are mostly occurred in areas which has precipitation greater than or equal to 3 mm/day.