Radial basis function neural network for sugarcane yield prediction

Crop yield prediction in terms of fresh weight is one of the many applications of machine learning in agriculture. Accurate yield prediction models are crucial as they guide growers in making appropriate decisions on what and when to cultivate under varying circumstances influenced by climatic and crop growth parameters, and market conditions. Production of sugarcane involves activities that heavily rely on accurate and timely cropping and harvest forecasting. This study aims to introduce a machine learning model, that estimates sugarcane yield based on various agronomic data collected in southwest Iran. Radial basis function neural network (RBF-NN), a simple shallow feedforward neural network that is distinguished for its simple structure, universal approximation, and fast learning speed, was employed to develop a yield prediction model. By utilizing datasets containing ۹ types of input variables and determining optimal values for network hyperparameters, including the number of hidden neurons and initial bandwidth value, different algorithms were examined for network training. The RBF-NN trained by the Levenberg-Marquardt algorithm, containing ۷۵ neurons in the hidden layer and a bandwidth value of ۰.۹, while using ۸۰% of the total data for training, achieved the highest accuracy, efficiency, and the least amount of estimation errors. Sensitivity analysis revealed that chemical fertilizers had the most significant impact on the yield estimation, while the opposite was true for electrical conductivity of the soil and month of the harvest.

کلیدواژه ها:

Artificial neural network (ANN) ، Harvest forecasting ، Machine learning ، Predictive modeling ، Smart farming