Utilizing IoT Monitoring and Machine Learning for Sustainable Agriculture Development

Abstract

In this extensive investigation, we introduce a method, an innovative agricultural monitoring system empowered by IoT technology, designed to meet the diverse requirements of farmers. This system integrates sensors to monitor moisture levels, regulate water pumps, and track temperature and humidity, ensuring a comprehensive approach to precision farming. The study explores the assessment of four distinct machine learning models: Random Forest, Support Vector Machine (SVM), Neural Network, and Decision Tree. Notably, the experiments encompassed a range of crops, such as wheat, rice, and soybeans, to evaluate the adaptability of the models across various agricultural contexts. Among the models examined, SVM emerges as the most promising candidate, demonstrating outstanding performance. Specifically, the SVM model with C=1.0 and 'rbf' kernel achieves an accuracy of 0.92, precision of 0.94, recall of 0.89, F1 score of 0.91, and ROC AUC of 0.95. These results underscore the potential of the novel method and machine learning to transform precision agriculture across different crops, providing a customized and data-driven approach to sustainable farming methods.