Using Forecasting Models to Optimize Production Schedule in a Cafe

Abstract

The purpose of the thesis was to research possible methods to predict the demand of the products and choose the best one to be used in the production schedule optimization module to be built in the cafe production management application, developed for the client. History production data from the client’s cafe was analyzed using R language. Three forecasting models were evaluated - ARIMA, support vector regression (SVR) and simple probabilistic model. Overall, the research shows that it can be feasible using examined forecasting models to predict product demand. The results were demonstrated to the client and the client sees the potential in the examined prediction algorithms. The SVR-based method showed good performance on all time series and produced particularly interesting results on one of them. The study shows that it will be worth integrating the SVR-based solution in the application in testing mode to verify its performance. It will also be worth carrying out deeper research as more data will be collected.