Machine Learning‐Based Hybrid Model for Wheat Yield Prediction

Abstract

Machine learning (ML) approaches are significant in resolving real-world issues. These methods have been used in the agricultural industry to help farmers anticipate crop yields, detect crop illnesses, and perform other tasks. In this chapter, a hybrid model for wheat crop yield prediction is presented, which combines feature selection with ML technique. The proposed hybrid model has two stages: the first uses a feature selection strategy to find the best features for wheat crops, and the second uses ML to estimate crop yield based on those best features. In this study, 12 different hybrid models have been implemented by integrating two feature selection techniques, each with six ML techniques in order to determine the optimum hybrid technique for wheat yield prediction. The applied feature selection techniques were genetic algorithm (GA) and ReliefF algorithm, while ML approaches used were K-nearest neighbor (KNN), Naïve Bayes, artificial neural network, logistic regression, support vector machine, and linear discriminant analysis. Five performance metrics, accuracy, recall, precision, f-score, and kappa coefficient, were used to assess the performance of these hybrid models. Among the 12 implemented hybrid approaches, GA-KNN model predicted the highest performance with accuracy of 98.3%.