Rainfall Prediction with Support Vector Machines: A Case Study in Tanjungpinang City, Indonesia

Abstract

Rainfall forecasting is becoming more challenging due to extreme climate change. Especially for the archipelago which has a unique geography compared to the mainland. The aim of this study is to test the performance of the support vector machine in predicting rainfall in Tanjungpinang, Kepulauan Riau, Indonesia. The variables used to predict are temperature, humidity, wind speed, and rainfall. The results obtained is a precision value of 82% for rain, with a ROC curve evaluation score of 0.74. These results show that the model built has a fairly good ability to separate between positive and negative results in predicting rainfall.