RAINFALL PREDICTION IN BALIKPAPAN CITY USING SUPPORT VECTOR REGRESSION

Abstract

Support Vector Regression (SVR) is one of the machine learning methods. The concept of SVR is to maximize the hyperplane to obtain support vector data. In machine learning, there is an overfitting problem, namely the behavior of the data during the training phase produces almost perfect accuracy. The advantage of the SVR method is that it can produce good predictions because it is able to solve the overfitting problem. The Central Statistics Agency stated that Balikpapan City experienced natural disasters such as landslides and floods due to high rainfall. This study aims to obtain the best SVR model with a linear kernel in predicting rainfall in Balikpapan City. The data for this study are monthly data on rainfall in Balikpapan City from January 2014 to December 2023. The proportion of training data and testing data used is 60:40, 70:30, 80:20, and 90:10. The results of the study indicate that the best SVR model with a linear kernel is a model with a proportion of training data and testing data of 90:10 where the parameters used are ε = 0,9 and C = 128 with a support vector of 24 points and a bias value of 0.02818046. The results of the Balikpapan City rainfall prediction are quite good, indicated by the RMSE value of the training data of 0.159 and the RMSE value of the testing data of 0.150.