Forecasting Patient Needs in a Donor Funded Health Care Project in Kenya

Abstract

Forecasts are crucial for practically all economic and business decisions. The focus of this research paper is in the area of forecasting. The research approach adopted is a case study of the Nutrition and HIV Program (NHP), which is a donor funded public health project. The general objective of this paper was to forecast the demand for patient needs in a donor funded project. Specifically, this paper sought to establish a suitable forecasting method that can accurately predict demand for nutrition commodities. In order to establish a more suitable forecasting method, Univariate Box – Jenkins (UBJ) methodology was used and two models were tested and Auto Regressive Integrated Moving Average (ARIMA (0, 1, 2)) model provided a better fit and was chosen as the model of choice for a short run forecast horizon. The main conclusion drawn from this paper is that, UBJ-ARIMA models are useful as benchmarks for forecasting and therefore
they should be viewed as complements to a reliable forecasting process. This paper recommends that public health projects need to consider adopting business forecasting methods that will provide a better glimpse of the future based on historical events rather than relying on disease morbidity data trends.

Key words: Autocorrelation function, ARIMA, partial autocorrelation function, public health project, residual autocorrelations short run forecast, stationarity, UBJ, un-differenced