Forecasting The Top Five Philippine Import and Export Commodities Using XGBoost and Hybrid Models

Authors: Marvin Tagaro Palabon

DOI: https://doi.org/10.37082/IJIRMPS.v14.i1.232954

Short DOI: https://doi.org/hbqgwb

Country: Philippines

Full-text Research PDF File:   View   |   Download

Abstract: This study forecasts the quarterly import and export performance of the Philippines at the commodity level for the period 2025–2030. Using quarterly data from 2005 to 2024, the study examines the historical behavior of the top five import and export commodities, compares the forecasting performance of different models, and generates medium-term trade projections. Extreme Gradient Boosting (XGBoost) is employed as the primary forecasting method to capture nonlinear and lagged dynamics in trade series, while traditional ARIMA and Hybrid ARIMA–XGBoost models are used as benchmark specifications for model comparison. Long-term trade trends are also assessed using Compound Annual Growth Rate (CAGR) analysis. The results show that machine learning–based models, particularly Univariate XGBoost, outperform traditional ARIMA models for most import and export series in terms of forecast accuracy. Hybrid models provide additional gains for selected commodities with more complex dynamics. The forecast indicates sustained demand for transport equipment and mineral fuels for imports, while electronics-related products—especially semiconductors and ignition wiring sets—remain dominant among export commodities.
Overall, the study demonstrates the effectiveness of combining econometric and machine learning approaches in forecasting international trade at the commodity level. The findings provide forward-looking and data-driven insights that can support trade monitoring, policy analysis, and strategic planning for the Philippine economy.

Keywords: Trade Forecasting, XGBoost, ARIMA, Imports, Exports

Paper Id: 232954

Published On: 2026-02-24

Published In: Volume 14, Issue 1, January-February 2026

Share this