Implementation of the Support Vector Regression (SVR) Method in Inflation Prediction in Makassar City

Abstract

Inflation is an important economic indicator, the growth rate is always kept low and stable. One step to deal with the possibility of a high inflation rate is to know the picture of the inflation rate in the future by making predictions. Prediction is a method used to determine a value or need in the next period. Support Vector Regression (SVR) is a development of the Support Vector Machine (SVM) method which is used for regression cases which can handle non-linear data cases. The problem that often occurs when using the SVR method is determining optimal model parameters. One way to determine the best parameters for the SVR method is to use Grid Search Optimization. The stages of the SVR method include data normalization, dividing training data and testing data, using the Radial Basis Function kernel, selecting the best parameters using Grid Search Optimization, and making predictions using the best model obtained with parameters γ = 10, ∁ = 100, and ε. = 0.1 with k = 5. The prediction results obtained were then evaluated by looking at the RMSE value, the RMSE value obtained was 0.029, which means the model's ability to follow the data pattern well and the prediction results made were very good.