Enhancing Sustainable Agriculture Through Smart Architecture: An Adaptive Neuro-Fuzzy Inference System with XGBoost Model

Precise agricultural production forecasting is difficult due to the deviations in environmental, soil, and crop conditions, as well as the complex interactions of affecting factors. Current methodologies, such Adaptive Neuro-Fuzzy Inference Systems (ANFIS) and Artificial Neural Networks (ANNs), have enhanced prediction efficacy but encounter obstacles such as substantial processing requirements, data reliance, and restricted generalisability. This study examines problems such as the necessity for enhanced model interpretability, scalability, and adaptation to varied agricultural situations. We present a hybrid ANFIS-based methodology optimised by the incorporation of sophisticated algorithms. The suggested system attained exceptional performance, evidenced by a coefficient of determination (R2) of 97.9%, a Root Mean Square Error (RMSE) of 0.017, and a Mean Absolute Percentage Error (MAPE) of 0.003, indicating its high precision and dependability. These indicators emphasise the system's capacity for accurate yield forecasting and effective decision-making in agricultural management.