An Ensemble Gazelle ANN and Extreme Gradient Boosting Technique for IoT Integrated Fertilizer and Manure Recommendation System Based on Soil Quality

Agriculture plays an important role in the Indian economy. Soil quality monitoring is critical for achieving profitable farming. Conventional soil quality testing requires a laboratory, which is costly and labor‐intensive. This also makes it unsuitable for continuous soil quality monitoring. Recent Internet of Things (IoT) and soil sensor developments are advancing technologies toward continuous monitoring of soil quality utilizing machine learning (ML) approaches. The low accuracy of the existing model can be caused by insufficient soil data, poor data quality, and ineffective feature selection. The proposed work aims to develop a soil quality monitoring system through the integration of unique ML approaches with IoT sensors. After that, the Node‐Micro Controller Unit (Node‐MCU) collects the data and stores it in the cloud. To determine soil quality, the gathered real‐time data is first pre‐processed by replacing missing values with the mean value and data normalization using log normalization, which enhances dataset quality and prediction accuracy. Then, optimal features are selected using the Ant colony Aquila Optimization algorithm (AcAOa). Finally, an Ensemble Gazelle Artificial Neural Network and the Extreme Gradient Boosting technique (EGANN‐XGBoost) were used to forecast soil quality and recommend the required fertilizer and manure to improve soil quality. Here, the Gazelle Optimization Algorithm (GOA) is utilized for hyperparameter tuning. The proposed method was evaluated using a variety of performance metrics, and it achieved 99.47% accuracy and 98.79% precision. The performance analysis demonstrates that the proposed model outperforms existing methods.