An Explainable Machine Learning Approach for Agricultural Crop Recommendation

Crop selection is a pivotal decision-making process for farmers, impacting cultivation outcomes and the efficient allocation of resources. Traditionally, crop selection has been based on farmers' experience and local knowledge. However, under dynamic environmental conditions such as soil quality, climate change, and weather variability, this traditional approach makes it difficult to make well-informed decisions. This paper introduces “Harvestify,” a crop recommendation system that combines predictive accuracy and interpretability. Harvestify in-corporates various machine learning (ML) techniques, including Random Forest, Support Vector Machines, K-Nearest Neighbors, LightGBM, and XGBoost, and rigorously evaluates their crop prediction capabilities. Comparative analysis revealed that the XGBoost classifier demonstrated superior performance, leading to its adoption as the implemented model. To ensure transparency, the SHAP (Shapley Additive explanations) approach is introduced to generate interpretable explanations for crop recommendations based on various soil and climate parameters. The SHAP approach is an interpretability framework based on game theory, assigning a unique value to each feature in the dataset, indicating the contribution of each feature to the model's prediction. This research demonstrates the synergistic integration of machine learning techniques and explainable artificial intelligence (AI) for precision agriculture, providing data-driven insights to farmers while maintaining transparency in decision-making.