AI-Driven Precision Agriculture for Smallholder Farmers in Rwanda: A Case Study in Kayonza District

Abstract

This research investigates the transformative role of Artificial Intelligence (AI)-driven precision agriculture in enhancing climate resilience, productivity, and financial protection for smallholder farmers in Rwanda, focusing on Kayonza District. The study integrates satellite-based vegetation indices (NDVI), rainfall anomaly datasets (CHIRPS, TAMSAT), IoT-enabled soil sensors, and machine learning algorithms, particularly Random Forest models, to improve crop yield prediction, drought monitoring, and agricultural risk management. Model evaluation demonstrated strong predictive capacity (R² = 0.83; RMSE = 1.21 t/ha), with soil moisture, NDVI, and rainfall anomalies identified as key yield determinants. SHAP analysis enhanced model transparency, informing actionable insights for tailored interventions. The study introduces and field-tests an AI-powered, revenue-index insurance model that bundles crop and livestock protection, credit access through savings cooperatives, mobile climate alerts, and digital extension services. It further explores innovative solutions such as drone-assisted irrigation in drought-prone zones like Ndego to address water scarcity with precision. Recommendations emphasize institutional integration of digital risk management tools into national agricultural systems, scaling mobile advisory services, expanding data-driven claim triggers, training agronomists on AI applications, and strengthening partnerships with cooperatives and insurers for holistic risk protection. The findings validate the scalability of AI-powered tools to inform digital agriculture policy, advance sustainable farming practices, and promote inclusive agri-financing. Overall, the study contributes a practical roadmap for leveraging modern technology to build climate-resilient food systems and improve rural livelihoods in Rwanda.