An Empirical Study on Food-Retail Demand Forecasting: A Comparative Analysis of Time Series Deep Learning Models

다국어 초록 (Multilingual Abstract)

Accurate demand forecasting in the food-retail industry is a very important task since it can reduce the cost caused by either shortage or overflow of food materials. In this paper, we show a comparative analysis on food-retail demand forecasting using the following two time series deep learning models: long short-term memory (LSTM) and convolutional neural network (CNN)-LSTM models. Using a café point-of-sale (POS) dataset, it is demonstrated that the CNN-LSTM model has a marginal gain over the LSTM model in terms of prediction error.