Binary vs Multi-class with Gaussian Filter on Typhoon Image Classification for Intensity Prediction

Strong meteorological events Tropical Cyclones (TCs) pose serious risks to coastal ecosystems and communities. Their strength is usually categorized using a variety of metrics, including wind speed, pressure, and rainfall since it directly corresponds with the possibility of damage and fatalities. An accurate classification of TC severity is essential for disaster preparedness, response plans, and mitigation initiatives. Support vector machines (SVM) {function category}, K-Nearest Neighbors (KNN) {lazy category}, Bayesian networks {Bayes category}, Random forests {Ensemble category}, and decision trees {Tree Category} are among the machine learning classifiers whose performances are compared in this study in binary and multi-class configurations by using Gaussian image processing technique. Performance measures, including time complexity, ROC, PRC, accuracy, precision, recall, and F-measure, were examined. The results indicate that Multi-class with SVM and Multi-class with Random Forest classifiers consistently outperform other models across most metrics, achieving the highest accuracy (0.88) and superior ROC (0.97) and PRC (0.94-0.95) scores. However, SVM models exhibited significantly higher time complexity, particularly in the multi-class with SVM.