Evaluating the Performance of Data-Driven Models Combined with IoT to Predict the Onion Yield under Different Irrigation Regimes

Abstract

Major international nutrition organizations are becoming increasingly concerned about global agriculture production. In particular, food insecurity has emerged in Egypt as a result of population expansion and rising food demand. Climate change and its variability are significant factors to the worldwide food insecurity problem. Therefore, cutting-edge techniques like IOT and machine learning which have shown its worth in several ways are needed by farmers and agricultural decision-makers to assist it in making timely judgments that will affect the quality of agricultural harvests. A novel open-source technology based on Arduino Board has been developed to predict onion crop irrigation needs. By utilizing three irrigation levels (100%, 85%, and 70% ETc) and integrating machine learning, the system predicts onion output in Talkha city, Egypt for 2023/2024. Meteorological and agronomic data are combined to aid farmers and decision-makers in predicting seasonal onion yields. Our system was constructed using Artificial Neural Network (ANN), Random Forest (RF), and Decision trees (DT) models. We found that, Using Arduino device at T85% and T70% helps to rationalize amount of water by 13% and 28% comparing with T100%. Furthermore, Arduino design with algorithm model gain a good value for (CWP) by (14.5 and 16 kg/m3) at T85% and T70%. Additionally, the results indicated that the ANN model works well during the testing phase, with a R2 of 94.3%, while the RF model and DT perform 90% and 86.1%, respectively. Our results showcase the effectiveness of the technology in enhancing agricultural decision-making and crop management.