Forecasting Import-Export Volume and Value using Support Vector Machines and Random Forests

Abstract

National trade plays a crucial role in a country's economy by meeting the demand for goods and services. Imports and exports are vital components that significantly influence economic growth. This study aims to forecast and analyse the volume and value of imports and exports in the coming years, thereby helping the country to prepare for and enhance its competitiveness on the international stage. To achieve this, the study employs Support Vector Machine (SVM) and Random Forest methods, which have gained popularity for their high efficiency in handling classification and regression problems. These methods are particularly effective in producing accurate forecasts, especially when dealing with seasonal variations in data. The research also explores the potential of combining these methods for even more robust predictions. Additionally, the study considers external factors, such as global market trends and policy changes, which can influence trade dynamics. The findings of this research demonstrate the ability to predict future import and export volumes and values accurately. By comparing the error values in the form of Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE) from the model, it is proven that the SVM model is highly effective for export-import forecasting. Lower error values indicate higher prediction accuracy, highlighting the model's reliability in capturing complex trade dynamics. The insights from this study can be instrumental for policymakers and businesses in strategising for future trade activities, allowing them to make informed decisions that align with anticipated market shifts.