Classification of Hotspots Causing Forest and Land Fires Using the Naive Bayes Algorithm

Abstract

Forest and land fires that occurred in Indonesia have caused many losses for the community. Forest fires generally occur in August and September, coinciding with the dry season in most parts of Indonesia. One indicator of the occurrence of forest fires is hotspots. This study uses one of the data mining techniques, namely classifying hotspots in Riau Province. This study used a dataset of forest fires in Pelalawan Regency from 2015 to 2019 using the Naïve Bayes algorithm. The hot spots to be analyzed consist of temperature, humidity, rainfall, wind speed, and class. The highest accuracy of the dataset of forest and land fires in 2019 is 96.95%. The classification method using the Naïve Bayes algorithm can be used to predict the emergence of hotspots in the future so that they can take preventive measures before forest and land fires occur.