Title : Explainable neural forecasting for small-sample time series: A SHAP-based analysis of dengue incidence in cebu city, Philippines

Accurate and explainable forecasting of dengue incidence is essential for effective early warning and public health response. This study develops a deep learning framework for monthly dengue forecasting in Cebu City using 120 months of historical case data and remotely sensed environmental covariates. Time series diagnostics reveal strong trend persistence (F_T = 0.548), nonlinear lag dependence, and a mean-reverting AR(1) structure (φ = 0.79), with lagged cross-correlation observed for land surface temperature at 4–6 months.

Five models were evaluated namely, AutoARIMA, NeuralProphet, NBEATSx, DeepAR, and DeepNPTS. DeepAR achieved the highest individual predictive accuracy (MAE = 18.00, sMAPE = 8.44%), while a median ensemble of all models substantially outperformed every individual forecast (MAE = 8.67, sMAPE = 4.27%). NBEATSx was selected as the primary interpretable framework, balancing strong performance (MAE = 35.67, sMAPE = 19.20%) with built-in interpretability via trend seasonal decomposition and SHAP-based explanations.

SHAP analysis reveals that the trend component dominates forecasts, while lagged dengue cases contribute significant but oscillatory effects reflecting nonlinear autoregressive behavior. Exogenous variables provide minimal short-term predictive gains at the monthly level. These findings indicate that monthly dengue dynamics are primarily driven by intrinsic temporal structure in this small-sample setting.
 

Virgilio D. Espina is a Filipino Russian Government Scholar studying Masters in Big Data Analytics and Artificial Intelligence at Novosibirsk State University, Russia. He also holds a Master’s in Applied Statistics from Mindanao State University – Iligan Institute of Technology, Philippines.