Novel BetaSutte for forecasting: Indonesian export prediction compared to random forest, XGBoost, and ETS

What is it about?

Our research introduces a new forecasting approach called "BetaSutte" and tests how well it predicts Indonesia's export values compared to established methods like Random Forest, XGBoost, and ETS models. Accurate export forecasting is crucial for businesses and policymakers in Indonesia, as exports significantly impact the country's economy. We analyzed monthly export data from 2022 to 2024—a period marked by global disruptions including the Russia-Ukraine conflict and post-pandemic economic shifts. Our findings show that while traditional models performed better at fitting historical data, the BetaSutte model was notably more accurate at predicting future export values. The BetaSutte approach works by separating long-term trends from short-term fluctuations, making it particularly effective during volatile economic periods. The model identified a systematic decline in Indonesia's export values, providing valuable insights for trade policy. This research offers a more reliable forecasting tool for businesses and government agencies involved in Indonesia's export sector, potentially improving decision-making in an increasingly unpredictable global trade environment.

Featured Image