Performance Evaluation and Prediction of Weather and Cyclone Categorization Using Hybrid Technique

Abstract

One of the greatest challenges faced by the meteorological department newlinearound the world is predicting the weather and severity of tropical cyclones. Even a newlinenation s or state s economy is affected by the climatology conditions which is more newlineimportant to predict, as they influence our daily life. Even after the advancement in newlineScience and Technology, the weather prediction accuracy is not adequate. Even today, newlinethe prediction remains as a major research topic for several researchers and scientists newlineintended to develop an algorithm or model which could help in prediction of weather newlineaccurately. Data mining with Machine Learning is the current technology used by many newlineresearchers for various applications. newlineIn the first phase of this research work, statistical analysis of weather data newlinecollected from SRM Automatic Weather Station, is done by applying data mining newlinetechniques such as C4.5, Random Forest and Naive bayes, which derive or extract some newlinerules that are used to predict weather.From the results, it is noted that the performance newlineof RF Decision Tree algorithm is better when compared with Naïve Bayes and C4.5 by newlineconsidering the precision values. newlineThe most endangered regions of cyclones formed in the world are Indian newlinesub-continent. The coastal line present in this region is about a total of 7516 km that newlineincludes 132 km in Lakshadweep, 5400 km main land, and Andaman and Nicobar newlineIslands includes about 1900 km newline