Tropical Cyclone Intensity Prediction Using Best Track Data Over North Indian Ocean By Machine Learning Classifiers

Intensity prediction of Tropical Cyclone is challenging and requires computational support to increase efficiency promptly. Best track data of tropical cyclones are used for intensity prediction. This study compared the correctly classified instances received by Naive Bayes, Logistic regression, Multilayer perceptron, Sequential minimal optimization, C4.5 decision tree, Random Trees, and Random forests machine learning-based classifiers. Five predictors: latitude and longitude, Central pressure and Pressure drop, and Maximum sustained wind speed over the North Indian Ocean (NIO) are used for classification. Best track data of Tropical Cyclones (TCs) from 2011 to 2020 are used for intensity prediction and comparison between various machine learning classifiers. In this study, we found that the classification accuracy reaches 97─99% with ML classifiers.