SMART INTEGRATED AGRICULTURAL MONITORING SYSTEM (SIAMS): An IoT and Machine Learning Approach for Crop Health and Irrigation Decision Support

The Smart Integrated Agricultural Monitoring System (SIAMS) is an IoT- and AI-powered solution aimed at optimising agricultural practices through real-time monitoring and predictive analytics. It involves converting low-cost sensor data into timely, farmer-friendly recommendations. I deployed an ESP32-based node at four farm sites in Osun, Ogun, Unilag, and Ikorodu, streaming readings to a cloud storage and transforming them into lagged and rolling features. Classical ML models (e.g., Random Forest/Gradient Boosting) forecast short-horizon soil moisture and classify dryness states, while a lightweight LLM layer turns the model outputs into structured, plain-English recommendations with confidence and rationale. A web dashboard (Streamlit on Azure) visualises trends and insights. On held-out data, SIAMS achieved RMSE = 0.7169, MAE = 0.4955, and R2 = 0.9979 with the Gradient Boosting ML model; ablations show lag features are critical. Field feedback indicates the recommendations were generally clear and actionable.This report presents the conceptual framework, system design, and implementation plans for SIAMS. It highlights the relevance of smart farming in addressing key agricultural challenges, particularly among smallholder farmers in Nigeria. The report also discusses the challenges, system evaluation strategies, and future directions for development, with emphasis on SIAMS’s potential to contribute to climate resilience, food security, and digital agricultural transformation.