Analisis Kinerja Metode Fuzzy Mamdani dalam Meramalkan Intensitas Curah Hujan di Kota Medan

Abstract

Indonesia is a tropical country with two main seasons, namely the dry season and the rainy season. Rainfall in a region is influenced by various meteorological factors such as air temperature, humidity, duration of sunshine, and wind speed. In an effort to predict rainfall more accurately, the fuzzy logic method is used because of its ability to handle uncertain and linguistic data. This study implements the Mamdani fuzzy method to predict monthly rainfall in the city of Medan based on four main parameters obtained from observational data from the BMKG Region I Medan. The prediction process consists of the following stages: fuzzification, application of fuzzy inference rules, and defuzzification. The system implementation was conducted over the period from January 2022 to July 2025. The system evaluation results using the Mean Absolute Percentage Error (MAPE) showed an error value of 43.06%, which is categorized as fairly good. This indicates that the system is capable of providing relevant rainfall estimates compared to actual data despite weather fluctuations. Based on these results, it is recommended that future research develop the system by adding additional variables such as air pressure or the ENSO index, as well as increasing the amount of historical data to obtain more optimal prediction results.