Disseminating the Process of Hurricane Path Prediction using Multilayer Perceptron and Support Vector Machine upon Varied Kernel Functions

Hurricane path prediction entails the prediction of future path data from a given path history. Path data, being an example of time series data, is a series of events organized chronologically. Presently, there exist a wide variety of algorithms at the disposal of data scientists to perform regressive prediction. This paper aims to evaluate and compare the efficacy of Multilayer Perceptron and Support Vector Machine at the task of Hurricane path prediction.

I. Introduction

A hurricane is a type of low-pressure weather system that is characterized by strong winds, heavy rain, and a low-pressure centre. These storms form over warm, tropical oceans and are fueled by the release of latent heat as moist air rises and condenses into clouds. As the storm system rotates and gathers strength, it can produce wind speeds greater than 119 km/h, with gusts even stronger.