Rwanda’s agriculture sector plays a critical role in the country’s economy, but farmers often face challenges in selecting the most suitable crops for their fields due to recommendation practices that focus primarily on soil nutrient analysis and neglect important factors such as climate variability and geographic altitude. This study presents the development of a data-driven crop recommendation system designed to address these limitations by using various data sources including soil test results, historical and forecast weather patterns, and crop-specific requirements, with machine learning algorithms applied to generate customized crop recommendations. While XGBoost achieved the highest accuracy at 98.25%, the Random Forest model, which achieved an accuracy of 96.7%, was selected for deployment because its better balance between precision and recall helps minimize the risk of resource inefficiencies from incorrect recommendations, and its more evenly distributed feature importance enhances generalizability across diverse farming conditions. The final model was integrated into a scalable analytics platform to ensure usability and impact, demonstrating how combining advanced data analytics with machine learning can enhance decision-making in agriculture, support sustainable farming practices, and improve yields in Rwanda.