Precision agriculture plays a critical role in addressing global food security while minimizing environmental impact. However, conventional fertilization practices often rely on fixed schedules without real-time feedback from soil conditions, leading to inefficient resource use and reduced crop quality. Unlike existing systems that provide static fertilizer schedules or lack integration with real-time soil data, UG-AgroPlan uniquely combines a calibrated multi-parameter soil sensor and a K-Nearest Neighbors model to deliver adaptive recommendations that dynamically adjust to changing soil conditions. This study introduces the UG-AgroPlan system, which integrates IoT-based soil nutrient monitoring with Machine Learning algorithms to provide real-time crop and fertilizer recommendations. The system utilizes a calibrated multi-parameter soil sensor capable of detecting nitrogen (N), phosphorus (P), potassium (K), pH, moisture, temperature, and electrical conductivity with high accuracy (97.41% Field validation was conducted using Uzbekistan melon as a case study, as this crop is highly sensitive to soil nutrient balance and requires precise fertilization to achieve optimal quality and specific location. This characteristic makes it an ideal indicator for evaluating the system’s accuracy and effectiveness, applying four fertilization strategies (daily, weekly, monthly, and conventional). Results showed that the daily precision fertilization strategy achieved the best performance, with fruit sweetness reaching 15.59°Brix, average weight 3.69 kg, and length 30.14 cm, while reducing fertilizer usage by up to 18% compared to conventional methods. These findings demonstrate that UG-AgroPlan offers a scalable, accurate, and adaptive approach to sustainable smart farming by integrating real-time sensing, machine learning, and actionable agronomic recommendations. These findings demonstrate that UG-AgroPlan offers a scalable, accurate, and adaptive approach to sustainable smart farming by integrating real-time sensing, machine learning, and actionable agronomic recommendations. Unlike existing systems that provide static fertilizer schedules or lack integration with real-time soil data, UG-AgroPlan uniquely combines a calibrated multi-parameter soil sensor and a KNN model to deliver adaptive recommendations that dynamically adjust to changing soil conditions.
Sugeru et al. (Wed,) studied this question.