Flood Prediction in Jakarta Using Long Short-Term Memory

Indonesia experiences a tropical climate characterized by two seasons: dry and rainy. Heavy rainfall during the rainy season can result in significant floods. Jakarta, particularly vulnerable to flooding due to intense rainfall and poor drainage infrastructure, presents an intriguing case for flood prediction. This study aims to forecast floods in Jakarta using a Long Short-Term Memory (LSTM) algorithm based on time series analysis. Weather data utilized includes temperature, humidity, and rainfall spanning from 2013 to 2023, totaling 4017 data points sourced from the Indonesian Meteorology, Climatology, and Geophysics Agency (BMKG). Weather forecasting is conducted using the API provided by Weather Visual crossing for the next 10 years. These weather predictions are employed to assess the likelihood of floods. The research findings demonstrate that the LSTM model achieves a 98% accuracy rate in predicting floods based on the weather prediction data.