KNN and C4.5 algorithms for predicting whirlwind disasters

Based on data taken from BNPB Indonesia, the incidence of whirlwind disasters has increased significantly in the last 10 years. In 2007, there were only 127 incidents in a year, but in 2017 there were 645 incidents. The losses due to the whirlwind disaster were not small, not only material losses but also recorded hundreds of people were injured and even dozens died. Therefore, the authors try to classify the weather when a whirlwind occurs using the KNN and C4.5 algorithms in the West Java region. The research objective was to provide predictions of whirlwind which often occurs in West Java. With a weather attribute approach taken from BMKG data and whirlwind disasters taken from BNPB Indonesia. This study is able to produce the best results with an accuracy of 83%, 85% precision, 96% recall and 0.691 AUC using KNN Algorithm.