Time Series Modelling and Forecasting of Birth and Death Rates: A Case Study of Grimard Catholic Hospital, Dekina LGA, Nigeria

Abstract

This study employs advanced time series modelling techniques to analyze and forecast monthly birth and death patterns recorded at Grimard Catholic Hospital, located in Dekina Local Government Area, Nigeria. Descriptive statistics reveal a stable trend in births with moderate seasonal variation, and a low, irregular pattern in mortality rates. Three univariate time series models: Seasonal-Trend decomposition using Loess (STL), AutoRegressive Integrated Moving Average (ARIMA), and Exponential Smoothing State Space (ETS), were applied to uncover temporal dynamics and project future values. Model performance was evaluated using Akaike- and Bayesian Information Criteria (AIC, BIC), alongside forecast accuracy metrics such as Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). Results show that ARIMA consistently outperformed both STL and ETS in forecasting births and deaths. The findings provide valuable insights for hospital resource planning and rural health policy development, emphasizing the critical role of statistical forecasting in informed healthcare decision-making.