Development of an Indonesian Trade Forecasting Information System Based on Statistical Models and Gradient Boosting

Current Indonesian trade forecasting relies on complex manual processes prone to inaccuracies. This study develops an Indonesian Trade Forecasting Information System integrating Statistical Models (SARIMA, Prophet) and Gradient Boosting (LightGBM, XGBoost, ExtraTrees). Using BPS data from 2012-2025, XGBoost achieves MAPE 18.64% for volatile exports while SARIMA records 7.37% for stable imports. TAM validation by 30 trade analysts shows high acceptance (PU=3.73, PEOU=3.82, BI=3.59). The system features interactive dashboards, secure authentication, and CSV/PDF exports, addressing national forecasting methodology gaps. Key contributions include dual-model integration for diverse trade patterns with user-friendly interfaces.