Evaluating the Forecasting Performance of Time Series Approaches on Measles Data in Adamawa State

Accurate forecasting of measles incidence is crucial for optimizing vaccination campaigns and strengthening disease control efforts in Adamawa State, Nigeria. This study undertakes a comparative evaluation of multiple time series models to determine their relative performances in predicting measles cases. Monthly measles incidence data spanning 2020 to 2024 were analyzed using Autoregressive Integrated Moving Average (ARIMA), Seasonal ARIMA (SARIMA), and Holt–Winters exponential smoothing models. Parameter estimation was carried out via maximum likelihood, and model adequacy was verified through residual diagnostics and Ljung–Box tests. Comparative evaluation employed the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Root Mean Square Error (RMSE) to assess in-sample fit and out-of-sample forecast accuracy. The Holt–Winters model achieved superior performance, yielding the lowest RMSE, AIC, and BIC values, followed by SARIMA (2,1,1)(0,1,1)12_{12}12 and SARIMA (1,1,1)(0,1,1)12_{12}12. These results demonstrate the effectiveness of exponential smoothing in capturing both seasonal and trend components of measles dynamics in the state. The findings provide an evidence-based modeling framework to support public health decision-making, enabling more proactive epidemic preparedness and targeted intervention strategies.