Comparison of Machine Learning Algorithms for Rainfall Prediction in Semarang City

Floods often occur in large urban areas and lowlands such as Jakarta, Semarang, Surabaya, etc. Based on disaster recapitulation data from Badan Penanggulangan Bencana Daerah (BPBD) / the Regional Disaster Management Agency of Semarang City in 2020, there were 23 floods in Semarang City. Based on the explanation, it is necessary to have an application that can predict rain in Semarang City to provide information about tomorrow’s rain prediction based on daily climate data from the BMKG Semarang Climatology Station. This study uses several machine learning classification methods to get the best model for predicting rainfall. In assessing the performance of the prediction model, several evaluations were used, such as accuracy, precision, recall, f-measure, AUC score, and 10-fold cross-validation. The results showed that the CatBoost Classifier prediction model had the best performance with an accuracy value of 76.77% and a precision value of 78.70%.