A Comparative Study of ARIMA, ARFIMA, and ARIMA-ARFIMA Models in Predicting Under-five Mortality Rate in Four East-African Nations

Despite significant advancements over the previous three decades, under-five mortality is still a significant public health concern in East Africa. Sustainable Development Goal (SDG) 3.2 calls for a reduction in under-five mortality to 25 deaths per 1,000 live births by 2030. Recent evaluations show that the area is not on course to attain the SDG objective, despite considerable declines in Kenya, Rwanda, Tanzania, and Uganda. This study compares the forecasting performance of Autoregressive Integrated Moving Average (ARIMA), Autoregressive Frictionally Integrated Moving Average (ARFIMA), and hybrid models for predicting under-five mortality rates (U5MR) in four East African countries and assesses their projected progress toward SDG 3.2.

Annual U5MR data for 1995–2022 were obtained from the World Bank. Differencing was used to attain stationarity after initial Augmented Dickey-Fuller (ADF) tests revealed non-stationarity in all four nations. ARIMA, ARFIMA, and hybrid models tailored to each country were fitted and assessed using AIC, BIC, RMSE, MAE, MAPE, and R². The Ljung-Box test was used to determine residual independence. The best-performing models were used to create forecasts for 2023 to 2030.

In every country, ARIMA models performed better than ARFIMA and hybrid models, exhibiting the best residual diagnostics and the lowest error metrics. Through 2030, U5MR is expected to continue declining, although none of the four nations are expected to meet the SDG 3.2 objective.

To achieve SDG 3.2 in East Africa, child survival initiatives and healthcare systems must be strengthened.

In every country, ARIMA models performed better than other models, showing the best residual diagnostics and the lowest error metrics. Although U5MR is expected to continue declining through 2030, none of the four nations is expected to meet the SDG 3.2 objective.