THE USE OF THE RANDOM FOREST ALGORITHM TO ANALYZE THE PREDICTION OF FOREST FIRES IN INDONESIA Anggelina Maria Putri Sihotang- Dhea Adelina Lumban Raja- Romasi Sihol Marito Siringo Ringo- Fitri Maya Puspita- Sisca Octarina
Program Studi Matematika Fakultas Matematika dan Ilmu Pengetahuan Alam Universitas Sriwijaya
Jalan Palembang-Prabumulih, Indralaya, Kabupaten Ogan Ilir (30662)
Abstract
Forest fires are one of the most devastating disasters worldwide, including in Indonesia. According to the Ministry of Environment and Forestry report, the total area affected by forest and land fires from 2015 to 2019 was 1.6 million hectares. Factors influencing forest fires include both natural and human-induced elements. Natural factors such as temperature, humidity, drought, El Nino, volcanic eruptions, and lightning play a role. Researchers believe that human activities in the forests, such as deforestation, timber exploitation, hunting, and slash-and-burn agriculture, contribute causally to forest fires, especially in large remaining forest areas. Despite some research on forest fire prevention, including the use of Data Mining and Machine Learning techniques to predict when forest fires might occur based on weather conditions and past fire reports, these efforts are not yet fully implemented. Therefore, in this study, we are developing a concept for a forest fire prediction system that can serve as a reference for policy-making when the government undertakes preventive actions. We are modeling annual forest fire data in Indonesia using the Random Forest Algorithm model, aiming to assist the government in preventing forest fires in accordance with legal guidelines. The analysis is also available at the Weather Modification Technology Center (BBTMC), aiding decisions on when weather modification may be necessary.
Keywords: Forest Fires- Data Mining- Random Forest
