Analysis of meal census patterns for forecasting menu item demand

Extract: Research was conducted to identify functional relationships between patient census and official institutional census recorded at midnight for hospital inpatients, and for mental health patients cared for in a satellite facility. Investigation of these relationships was necessary to provide knowledge essential for the design of a statistical menu item forecasting system. Patient census for 3 meals (breakfast, lunch, and supper) was compared with the midnight census, using graphical analysis and analysis of variance. Reliable patterns were identified in the patient census data which required accommodation in a mathematical forecasting model. Three forecasting design options were identified and appraised. The simplest design option and the 2 more complex design options forecasted identical quantities. (author)