LASSO Quantile Regression for Predictive Modeling of Dengue Hemorrhagic Fever Incidence in Indonesia

Arifin, Samsul (Department of Statistics, Lambung Mangkurat University, Banjarbaru)
Anggraini, Dewi (Department of Statistics, Lambung Mangkurat University, Banjarbaru)
Susanti, Dewi Sri (Department of Statistics, Lambung Mangkurat University, Banjarbaru)

Zero : Jurnal Sains, Matematika, dan Terapan
Vol 9, No 1 (2025): Zero: Jurnal Sains Matematika dan Terapan

Dengue Hemorrhagic Fever (DHF) is an endemic disease that continues to burden public health in Indonesia, characterized by an uneven pattern of distribution influenced by various environmental, social, and economic factors. This study aims to develop a predictive model for DHF incidence using the LASSO Quantile Regression approach, which can reveal the influence of predictor variables across different quantiles while addressing multicollinearity and overfitting issues. The data used includes nine predictor variables obtained from BPS and BMKG for the year 2025. The estimation results show that the urban/rural Area Size consistently affects all quantiles, while the percentage of population living in poverty and the number of healthcare facilities are significant only at the 0.25 and 0.50 quantiles. Model evaluation indicates that this approach provides good predictive performance, especially at the 0.25 quantile, with a R² pseudo value of 0.2838. These findings suggest that the LASSO Quantile Regression method is effective in identifying the determinants of DHF in Indonesia.