Klasifikasi Fire Weather Index untuk Indikator Kebakaran Hutan dengan Metode Support Vector Machine

Abstract

The Fire Weather Index is one of the main subsystems of the Canadian Forest Fire Danger Rating System (CFFDRS). The Fire Weather Index has been studied by several researchers for several geographic areas in the world and is proven to be an index for fire hazard assessment. The Fire Weather Index is based on the moisture content of three classes of forest fuels with the influence of wind on fire behavior. The FWI consists of six components: three primary sub-indices representing fuel moisture, two intermediate sub-indices representing the rate of spread and consumption of fuel, and a final index representing fire intensity as the level of energy output per unit length of the flame front. Accurate and precise weather observations that meet the specified standards and specifications required for accurate and representative calculations of all components of the Fire Weather Index. Data mining is finding interesting patterns or new information from large amounts of data. Classification is a technique in the field of data mining that is used to form predictive models for data classes. The method used is the Support Vector Machine (SVM) method. The SVM method uses the kernel in a nonlinear mapping to convert the original training data to a higher dimension. The kernel used in this study is the Radial Basic Function (RBF) kernel. System training and testing are carried out by measuring the results of accuracy, precision, recall, and f1-score. This study aims to determine the class categories of the Fire Weather Index, namely low, moderate, high and extreme, using the Support Vector Machine method in the North Sumatra region.