Developing an Optimal Model for Predicting the Severity of Wheat Stem Rust (Case study of Arsi and Bale Zone)
This work addresses a domain-specific problem for agricultural researchers and farmers by providing an incremental improvement in predictive modeling for wheat stem rust.
This research tackled predicting wheat stem rust severity using three artificial neural network models, finding that the General Regression Neural Network (GRNN) was most effective with less training time and that total seasonal rainfall positively influenced rust development.
This research utilized three types of artificial neural network (ANN) methodologies, namely Backpropagation Neural Network (BPNN) with varied training, transfer, divide, and learning functions; Radial Basis Function Neural Network (RBFNN); and General Regression Neural Network (GRNN), to forecast the severity of stem rust. It considered parameters such as mean maximum temperature, mean minimum temperature, mean rainfall, mean average temperature, mean relative humidity, and different wheat varieties. The statistical analysis revealed that GRNN demonstrated effective predictive capability and required less training time compared to the other models. Additionally, the results indicated that total seasonal rainfall positively influenced the development of wheat stem rust. Keywords: Wheat stem rust, Back propagation neural network, Radial Basis Function Neural Network, General Regression Neural Network.