RB-Pred

RB-Pred: SVM-based prediction of rice blast severity using weather variables

RB-Pred predicts rice blast severity from six significant weather variables using a Support Vector Machine (SVM) model to forecast disease outcomes across temporal and geographic datasets.

Key Features:

• SVM Prediction Model: Implements Support Vector Machine algorithms to predict rice blast severity from six selected weather-related variables.
• Five-Fold Cross-Validation: Applies five-fold cross-validation for cross-year and cross-location model development and validation.
• Performance Benchmarking: Achieves cross-year performance of r = 0.77 and %MAE = 36.66, outperforming multiple regression (REG; r = 0.50), back-propagation neural network (BPNN; r = 0.60, %MAE = 52.24), and generalized regression neural network (GRNN; r = 0.70, %MAE = 46.30); achieves cross-location performance of r = 0.74 and %MAE = 44.12, exceeding REG (r = 0.48), BPNN (r = 0.56), and GRNN (r = 0.66).

Scientific Applications:

• Disease Forecasting: Quantifies rice blast severity under varying environmental conditions to support plant disease epidemiology and crop management strategies.

Methodology:

Selects six weather variables associated with rice blast severity and trains SVM models to predict disease outcomes. Validates models using five-fold cross-validation across cross-year and cross-location datasets to assess generalizability and predictive accuracy.