A Comparative Study of Ensemble Classification Algorithms for Crop Yield Forecasting

Abstract

Normias Matsikira, Gideon Mazambani and Martin Muduva

This study explores the application of ensemble learning techniques to improve predictive model accuracy. It focuses on combining classifiers to outperform individual models using structured and unstructured data from agricultural datasets. Artificial neural networks (ANNs) and ensemble methods were used to increase deep neural network efficiency. Experiments with different network structures, training iterations, and topologies were conducted, evaluating measures like sensitivity and specificity. The research also includes predicting crop yields using ensemble classification algorithms, comparing accuracy with conventional methods. The study highlights the importance of crop yield prediction for agricultural management and discusses the benefits of ensemble methods. Results show that Random Forest, XGBoost, AdaBoost, and ANNs perform well in predicting crop yields as compared to the other algorithms. This research contributes to understanding the impact of weather patterns and genotype on crop yields.

PDF