Machine Learning-Based Dual Agricultural Decision Support System: An Integrated Approach for Crop and Fertilizer Recommendations

This study proposes a comprehensive dual-module agricultural decision support system that integrates machine learning algorithms for both crop and fertilizer recommendations. The system addresses critical challenges in modern agriculture by providing evidence-based recommendations to optimize productivity while maintaining sustainability. Four machine learning algorithms (Random Forest, Support Vector Machine, XGBoost, and K-Nearest Neighbors) were implemented and evaluated using comprehensive agricultural datasets containing 2,200 crop samples and 3,100 fertilizer samples. The crop recommendation module achieved 99.32% accuracy using Random Forest, while the fertilizer recommendation module attained 98.75% accuracy. The system incorporates advanced techniques including SMOTE for handling class imbalance, GridSearchCV for hyperparameter optimization, and SHAP analysis for model interpretability. Comparative analysis with existing literature demonstrates competitive performance while offering enhanced explainability and dual functionality. The developed framework provides a foundation for region-specific implementations and represents a practical contribution to intelligent agricultural decision support systems.