Enhancing Agricultural Decision-Making through an Explainable AI-Based Crop Recommendation System

Agriculture is essential for maintaining food security and promoting economic sustainability; however, farmers frequently encounter difficulties in choosing the most appropriate crops for their land due to diverse environmental, soil, and market conditions. This paper proposes an innovative crop recommendation system powered by Explainable Artificial Intelligence (XAI), designed to enhance agricultural decision-making. The system utilizes machine-learning models to analyze diverse data inputs—such as soil properties, weather conditions, and market trends—to recommend optimal crops for specific regions. What sets this approach apart is its explainability: the XAI framework provides transparent insights into how and why certain recommendations are made, enabling farmers to trust and understand the decision-making process. The incorporation of eXplainable AI into crop recommendation systems has transformed agriculture, facilitating data-driven decision-making that leads to improved crop production and more efficient resource management. However, these models’ lack of transparency and interpretability frequently constrains their practical implementation. We explore XAI based approaches, such as LIME and SHAP, to interpret model outputs and highlight key features influencing predictions. Our experiments demonstrate that XAI improves the transparency of ML models and aids in refining model performance through informed feature selection and model adjustments. Our proposed models achieve the highest accuracy $\mathbf{9 9. 3 9 \%}$ using the KNN algorithm and BiLSTM model achieve the $\mathbf{9 5. 9 1 \%}$ accuracy.