Application of Support Vector Machine on Drought Code Classification in North Sumatra Indonesia

Abstract

This study aims to classify the Drought Code in North Sumatra. Drought Code is part of Fire Danger Rating System (FDRS) developed in Canada. One of the sub-systems of FDRS is Fire Weather Index (FWI) that aims to evaluate fire hazards from current and past weather conditions. Drought Code is classified by using Support Vector Machine. Support Vector Machine is widely used in the data classification process. One of the advantages of Support Vector Machine methods is it has ability in classifying large amount of data and classifying more than two classes or multi-classes. Weather parameters used in this study are rainfall and temperature in North Sumatra. The data used are from 8 (eight) meteorological observation stations in North Sumatra from 2017 to 2021. Drought Code is carried out with several tests using several kernels contained in SVM.