Using Holt Winter’s Multiplicative Model to Forecast Assisted Childbirths at the Teaching Hospital in Ashanti Region, Ghana

Abstract

The use of maternal healthcare facilities is an important indicator of the impact of the free maternal healthcare policy aimed at improving health status of pregnant women in Ghana. This study investigated the pattern of quarterly assisted deliveries at Komfo Anokye Teaching Hospital (KATH), Kumasi, Ghana from 2000 to 2011. The Holt Winters multiplicative and additive forecasting models were considered. The Multiplicative model reported a Root Mean Square Error of Prediction (RMSEP) of 31.10, Root Mean Square Error (RMSE) of 188.080, Mean Absolute Percentage Error (MAPE) of 6.2951 and Mean Absolute Scaled Error (MASE) of 0.7086 while the additive model reported RMSEP, RMSE, MAPE and MASE of 49.59 201.83, 6.3098 and 0.7106 respectively .The multiplicative model further passed the Shapiro-Wilks test (p-value 0.07358). Results identified the second and fourth quarters as peak seasons and the first quarters as deep seasons for assisted childbirths in the hospital.  The negative binomial regression confirmed this by identifying April, May, October and November as peaks months with May being the most significant month.

Keywords: Holt Winter’s multiplicative and addictive models, Forecasting, Seasonal, Negative binomial model.