Geometrically Innovated Machine Learning for Optimized Prediction of Rice Blast Disease

Rice blast disease, caused by the fungus Magnaporthe oryzae, is a major threat to global rice production, leading to significant economic losses and impacting food security. Predicting outbreaks of this disease is crucial for timely intervention and management. This paper introduces a novel machine learning technique, the Geometrically Innovated Machine Learning (GIML) method, which leverages specific agronomic parameters to enhance prediction accuracy. The study focuses on optimizing the parameters Temperature, Humidity, Soil Moisture, Soil pH, Nitrogen (N), Phosphorus (P), and Potassium (K), demonstrating their critical role in disease prediction. The GIML method integrates geometric feature engineering by calculating angles and magnitudes between data vectors, capturing intricate relationships among parameters that traditional models may overlook. Comprehensive experiments comparing this technique with conventional methods show significant improvements in prediction metrics. The results suggests that GIML outperforms the conventional methods like Support Vector Machine(SVM), K-Nearest Neighbour(KNN) and Decision Trees(DT) not only enhances model performance but also provides insights into the most impact parameters, paving the way for advanced predictive analytics in agriculture. This research underscores the potential of geometric feature engineering in transforming disease prediction models and suggests avenues for future work in other agricultural contexts.