AN IOT–MACHINE LEARNING–DECISION SUPPORT SYSTEM FRAMEWORK FOR SMART AGRICULTURE: DESIGN, IMPLEMENTATION, AND PERFORMANCE EVALUATION

Abstract

Smart agriculture integrates sensing technology, data analytics, and intelligent automation to address increasing global demands for food production, resource optimization, and sustainable farming practices. This research presents a comprehensive IoT–Machine Learning–Decision Support System (IoT–ML–DSS) framework designed to monitor crop health, optimize irrigation, and support real-time farm management decisions. The system incorporates a hierarchical IoT architecture, a multi-model machine learning pipeline, and a rule-based DSS for actionable recommendations. A custom multi-season dataset was developed to train and validate the system. Experimental results demonstrate high performance across disease classification (98.12% accuracy using CNN), irrigation prediction (MAE = 0.33), soil moisture estimation (R² = 0.94), and predictive maintenance of irrigation pumps (94.7% anomaly detection accuracy). Field deployment further shows reduction in water consumption by 23%, early disease detection improvement by 31%, and overall productivity gains of 18%. The findings indicate that the proposed IoT–ML–DSS framework offers a scalable, cost-effective, and reliable solution for precision agriculture.