Demand forecasting of meal in university dining services with short supply

Abstract

This paper aims to examine the meal demand forecasting in a University Dining Service (UDS) with short supply. The research derived from low productive capacity problems faced in some campus of the São Paulo State University (UNESP), which do not meet all demand. To estimate the proportion of people truly interested in the dining services and to calculate the surplus of non-service, it was suggested a design covering a combination of statistical techniques such as multiple regression analysis, diagnostic tests measurements, ROC curve, supported by a market research with quantitative approach.  With the utilization of these techniques combination, it was analyzed information based on socioeconomic profiles, menu requirements, reason for eating in the UDS, and the food habits of 544 academic people. After analysis, it was estimated a surplus of 311 daily non-services (78% over than offer). Most UDS users are undergraduate students in vulnerable financial conditions for food and residence; which use the UDS because of price; living near the campus; and are moderately demanding about the menu. The conclusions reinforce the relevance of contextual information about the service user in the demand forecasting model aiming to increase the estimate accuracy of quantity of non-service.