Penentuan Derajat Kerawanan Kebakaran Hutan Menggunakan Algoritme C4.5 dan M5P (Studi Kasus : Data Kebakaran Hutan Sumatera Selatan 2004)

Abstract

South Sumatera is one of Indonesia's region which has higher forest fire degree than other regions. Many factors indeed caused forest fire became wider for this region such as transmigration, illegal logging, etc. For that reason, the government created SSFFMP (South Sumatera Forest Fire Management Project) that used as FTA (Fire Threat Analysis). In SSFFMP's system, there is a fonnula to predict the forest fire degree using simple weight for three attributes such as landcover, peat and elevation. Therefore. the research tried to analyze the influence of other spatial attributes such as mainroad distance, river distance, and residence distance that was applied in the prior formula. The technique which is used for this research is classification using C4.5 and M5P algorithm by calculating distance for other spatial attributes such as mainroad distance, river distance,and residence distance. From this analysis, it would give new set of rules or fonnulas to predict the forest fire degree by using C4.5 algorithm classification and M5P algorithm classification. The result of this research showed that elevation, kind of Landcover, mainroad distance and residence distance are the main factors that used to predict the forest fire degree of a hotspot in South Sumatera. The result also showed that classification methode with C4.5 algorithm and M5P algorithm give the result more inaccurate than the fonnula from SSFFMP in order to predict the forest fire degree in South Sumatera. Thus, classification methode with C4.S algorithm and MSP algorithm could not implemented practically in real condition whereas the SSFFMP formula could give better result with higher accuracy.