Comparative Analysis of Forecasting Models for Infant Mortality Rate in Somalia

Abstract

One of the major challenges to infants' survival is environmental or socio-economic, influenced by technological factors. This research bridges a gap in the area of infant mortality, where not enough research has taken place; the country, Somalia differs in its challenges compared to other countries worldwide. The aim of this study is to develop an efficient forecasting model using the Holt-Winters (H-W) method, which is very flexible and efficient in handling seasonal variation in time-series data. The historical data ranges from 2001 to 2023, showing the environmental and healthcare variables affecting infant mortality. This research compares the performance of the H-W model against other time-series models like autoregressive integrated moving average and Exponential Smoothing, and various machine learning algorithms. Based on seasonal trends captured by results, the H-W model forecasts a decline in the infant mortality rate for the next decade. Evidence of this is seen in the dropping mortality rate among males from 0.085 in 2024 to 0.064 by 2033, and that of the females also drops from 0.072 to 0.053 over the same period.