Monthly Rainfall Prediction Using the Facebook Prophet Model for Flood Mitigation in Central Jakarta

• Main Authors: Sulasikin, Andi, Nugraha, Yudhistira, Kanggrawan, Juan Intan, Suherman, Alex L.
• Format: Conference Proceeding
• Language: English
• Subjects: Facebook Prophet; Fitting; flood mitigation; forecasting; Government; Jakarta; Predictive models; rainfall; Social networking (online); Time series analysis; Urban areas
• ISSN: 2640-0545
• DOI: 10.1109/ICISS53185.2021.9532507

Jakarta has been known as the city where floods are prevalent. As the vital region in Jakarta where the center of government and business are located, Central Jakarta is inseparable from the flood when the rainfall is remarkably high. Therefore, the Jakarta Provincial Government need a data-driven policy to facing potential flood that may occur each year to protect the citizen from the threat of flood disaster. Monthly rainfall prediction can be a reference to determine the possibility of considerable loss and damage due to disaster threats. However, at this moment, it is still challenging to find a fitting forecasting model for this context. This paper reports a comparison of three different time series models: Seasonal Autoregressive Integrated Moving Average (SARIMA), Facebook Prophet, and Long Short-Term Memory (LSTM) to forecast monthly rainfall in Central Jakarta for up to two consecutive years. The result indicates that Facebook Prophet, with the lowest Mean Squared Error (MSE) and Root Mean Squared Error (RMSE), is the fittest model to predict the monthly rainfall in Central Jakarta. It shows that a high amount of rainfall will be seen in January and February 2021, which suggests we need to be prepared to anticipate the potential flood. Facebook Prophet shows promising results in supporting data-driven policy for flood mitigation in Jakarta. The development of this model in the future can be used as a baseline study to formulate a data-driven policy for flood mitigation in Jakarta.