PERBANDINGAN MODEL MACHINE LEARNING PADA KLASIFIKASI CURAH HUJAN DI BOGOR

Accurate rainfall prediction remains a significant challenge due to the involvement of complex physical processes and its substantial impact on various sectors of society. Rainfall prediction can be performed using classification techniques in Data Mining. Each algorithm employed for rainfall prediction may yield different performance outcomes, depending on factors such as the size of the dataset, the number of missing values, and the meteorological parameters utilized in the study. Selecting the appropriate algorithm for rainfall prediction continues to pose a challenge. This study aims to compare the performance of Naïve Bayes, Decision Tree, and Random Forest in order to identify the best model for classifying rainfall in Bogor Regency. The data utilized in this study includes maximum temperature, minimum temperature, average temperature, average humidity, duration of sunlight exposure, maximum wind speed, average wind speed, maximum wind direction, and rainfall. The dataset spans five years comprising a total 1.825 of data obtained from the Class III Citeko Meteorological Station. The results indicate that Random Forest, when trained with a smaller proportion of data compared to the proportion of test data to be predicted, achieves the best performance, with a precision of 59.1%, recall of 64.3%, and f1-score of 65.5%. This performance is attributed to the ensemble principle employed by Random Forest, which combines multiple weak learner trees to produce a robust learner tree.