Reducing food waste

: demand forecasting in catering services using data analytics

Abstract

Fluctuations and unexpected events in the search for meals translate into problems for organizations, leading to costs and/or food waste. Currently, canteen managers tend to predict the demand for a particular menu, mainly through intuition, and this forecasting approach may lead to overestimation or underestimation. So, in order to solve these problems and accurately predict the demand for a specific menu on a specific date, this project proposes the development of a forecasting system. In order to forecast demand, several factors were considered, such as the school calendar, weather conditions and special events. This study is supported by daily data corresponding to two years of activity of a university canteen. Ten causal and time series forecasting methods were applied to the data with the support of the R software. The Random Forests method demonstrated to have the best performance in estimating demand.

Date of Award	20 Jul 2020
Original language	Portuguese
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Vera Miguéis (Supervisor)