THE APPLICATION OF SUPPORT VECTOR MACHINES IN FORECASTING INDONESIA'S EXPORT VALUES

Abstract

Exports play a crucial role in Indonesia's economic growth, but fluctuations in export values can impact national economic stability. While there is existing research on export forecasting, the application of advanced machine learning methods such as Support Vector Machine (SVM) is limited. This study aims to forecast Indonesia’s export values using SVM based on monthly data from January 2017 to February 2025. The data were split into 80:20 proportions for training and testing, with input variables optimized using Partial Autocorrelation Function (PACF) analysis. Fifteen input schemes were tested, and the combination of lag 1 and lag 2 produced the lowest Mean Absolute Percentage Error (MAPE) of 5.04% on the test data, indicating very high accuracy. The forecasted results show a declining trend in export values from 21.87 billion USD in March 2025 to 20.66 billion USD in December 2025, driven by external factors such as global economic slowdown and commodity price fluctuations. Despite the decline, Indonesia’s export values remain relatively high compared to pre-2021 periods. This research highlights the effectiveness of SVM for export forecasting and suggests that this method could be used to inform policy decisions to mitigate global trade risks. Future research could explore the inclusion of additional external variables and other machine learning techniques to further improve forecast accuracy. The novelty of this study lies in the application of SVM for forecasting Indonesia’s export values, filling a gap in the literature on export forecasting models.