Time Series Modeling of Dengue Outbreaks Through Singular Spectrum Analysis Incorporating Lunar and Solar Calendars for Improved Forecasting

Abstract

Dengue Hemorrhagic Fever (DHF) is a tropical infectious disease transmitted by the Aedes aegypti mosquito and exhibits seasonal patterns with periodic increases in cases throughout the year. The control of vector-borne diseases such as DHF is very important for strengthening public health resilience against climate change, in line with the Sustainable Development Goals (SDGs) for Good Health, Well-being, and Climate Action. Therefore, this study was focused on Bogor city, which experiences high rainfall and continues to face an elevated risk of DHF. The objective was to develop a time series forecasting model to predict DHF outbreaks using Singular Spectrum Analysis (SSA). This is a statistical method for identifying patterns in time series data. Lunar and Solar calendars were adopted to capture seasonal patterns and determine the optimal window length for prediction. The results showed that the Lunar calendar more accurately captured local seasonal variation related to DHF risk. Moreover, the SSA model with one component and a window length of 7 achieved the best performance with a Mean Absolute Percentage Error (MAPE) of 0.0757. The forecast accuracy decreased with longer horizons, but the model provided reliable predictions for short-term periods (approximately 1 month, i.e., up to 4 weeks ahead), which were considered useful for planning DHF mitigation. The results emphasized that the combination of SSA with appropriate calendar systems could improve the accuracy of epidemiological predictions and support vector control policymaking in tropical regions.