Applying expert systems technology to the implementation of a forecasting model in foodservice

Abstract

The objective of the study was to develop a naive expert system model to replicate the knowledge, experience, creativity, judgment, and intuition of the forecast knowledge expert in foodservice. The knowledge-based system (expert system) is computer technology that guides the completion of tasks that usually require specialized knowledge and experience. One benefit of the technology is that an expert system makes it possible for non-experts to gain specialized assistance without dealing with the time and location limitations that often restrict human experts. These specialized systems gather and examine data to assist managers in making better decisions. The expert system model developed in this study served as a tool for the foodservice manager/dietitian in the forecasting process. A series of interviews with the knowledge expert (foodservice manager/dietitian) were made from August through December, 1993. Data for 110 menu items were collected and grouped according to student preference and food type. The collection site was a university dining center which utilized a food court concept. Three of the four service lines were forecast. These service lines included mexican, italian, classic, and deli menu options. A total of over 240,000 potential combinations of menu items exist based on the variety of offerings available to students. An expert system software shell and spreadsheet software were used to develop the system. A naive expert system for forecasting menu items was developed. It was then tested by the forecast knowledge expert/dietitian, the researcher and other users, to evaluate the user friendliness and accuracy. Evaluation included a six-week test period where the model was compared to the knowledge expert/dietitian forecast. Error was reduced, using Mean Absolute Deviation (MAD) and Mean Squared Error (MSE) from MAD of 22 and MSE of 718 to MAD of 0 and MSE of 1.