Rainfall Prediction based on Historical Weather Data using Naive Bayes Classification Model in Southeast Sulawesi

Abstract

Southeast Sulawesi is one of the provinces in Indonesia characterized by diverse topography and climate, making it challenging to accurately identify and predict rainfall patterns.
The aim of this study is to enhance our understanding of weather behavior in Southeast Sulawesi and provide a foundation for developing more advanced and region-specific weather prediction methods. The data used in this research consists of historical weather records obtained from the official BMKG (Meteorology, Climatology, and Geophysics Agency) website, containing features that significantly contribute to rainfall prediction. The method employed in this study is the Naive Bayes classification model, which involves several stages including data collection, pre-processing, and preparation for the modeling phase, ultimately generating rainfall prediction outputs. The results of the study yielded a rainfall prediction accuracy of 74.66%. For the rainfall class (0.0), the model achieved a precision of 82%, recall of 66%, and F1-score of 73%. Meanwhile, for the rainfall class (1.0), the model achieved a precision of 69%, recall of 84%, and F1-score of 76%. Despite some prediction errors, these findings indicate that the Naive Bayes method can serve as a solid foundation for the development of more sophisticated and tailored weather prediction models for the Southeast Sulawesi region.