F.Ti Ayyu Sayyidul Laily, Moh. Muzayyin Amrulloh, Didik Dwi Prasetya, Figo Kurniawan Siswanto
Accurate rainfall prediction is important for disaster mitigation, water management, and agricultural planning. Particularly, in areas with extreme monthly rainfall variations like Malang Regency, future weather prediction is a critical need. Classification of rainfall data could be considered to process information into valuable knowledge for the benefit of many people. This research utilized the Naive Bayes algorithm for classification, implemented within RapidMiner, using datasets from the Meteorology, Climatology, and Geophysics Agency (BMKG) collected over a one-year period, from January 2021 to February 2022. Several stages were needed to determine changes in rainfall patterns, consisting of data collection, understanding, preparation, modeling, evaluation, and dissemination. The dataset includes four parameters: average temperature, humidity, maximum wind speed, and wind direction. Three training and test data scenarios were implemented to optimize model performance. The study demonstrates that the Naive Bayes algorithm within RapidMiner effectively predicts rainfall categories, including light, medium, or heavy rainfall, with a maximum accuracy of 85.71%, making it a reliable tool for classifying rainfall intensity. © 2024 IEEE.
Universitas Negeri Malang, Department of Electrical Engineering and Informatics, Malang, Indonesia