<![CDATA[Agricultural exports are highly vulnerable to global price volatility and seasonal fluctuations, creating demand for more accurate forecasting methods. This study evaluates the Adaptive Neuro-Fuzzy Inference System (ANFIS) for forecasting Indonesia’s monthly agricultural exports, addressing a gap in the literature where soft computing approaches have rarely been systematically applied. Using official trade data from 2012 to 2025, two alternative training–testing schemes (75%:25% and 80%:20%) were implemented with standard preprocessing, and forecasting accuracy was measured using RMSE, MAE, and MAPE. The results show that ANFIS delivered accuracy within widely accepted thresholds under the 75%:25% split, while accuracy declined under the 80%:20% split. Theoretically, the study contributes by clarifying conditions for reliable neuro-fuzzy forecasting and emphasizing standardized evaluation protocols. Practically, the findings provide decision-relevant insights for policymakers and exporters, supporting export target setting, forward-contract planning during volatile price swings, and logistics coordination during peak harvest seasons.]]>
Copyrights © 2025
