Predicting Powdery Mildew on Grapes with Classification Algorithms by using Microclimatic Data from IoT Weather Stations

This paper presents a comparative analysis evaluating the performance of six classification algorithms in predicting grape powdery mildew. Utilizing daily climatic data collected from grapevines, the assessed models demonstrate high metric performance in predicting infections with the best model scoring ~0.92 f1-score during the testing phase. The findings significantly contribute to the domain of smart agriculture by expanding the utilization of existing IoT systems, which capture the climatic conditions of grapevines. This extension enriches the field of smart agriculture by enhancing predictive capabilities, offering valuable insights for agricultural practitioners and researchers aiming to optimize disease management strategies in agriculture.