Track: Global Business Management Education
Abstract
Support Vector Regression (SVR) is one of the methods used in the supervised learning process for regression cases that comes from the Support Vector Machine (SVM). SVR is a method to solve forecasting cases that can overcome overfitting so that it will produce a good performance and has an advantage in optimization with good generalizability and accuracy results. There are several choices of kernel functions that can be seen from its ability to work on the SVR method, one of the most popular and often considered the best is the Radial Basis Function (RBF) because based on some previous research this kernel shows the lowest error value compared to other kernels so that the purpose of this study is to compare the RBF kernel with other kernels to see the performance of the model and the accuracy of forecasting produced by using the kernel on time series data which has a seasonal pattern.