Title: Configuration and implementation of a daily artificial neural network-based forecasting system using real supermarket data
Authors: Ilham Slimani; Ilhame El Farissi; Said Achchab
Addresses: Al-Qualsadi Research and Development Team, National Higher School for Computer Science and System Analysis (ENSIAS), Mohammed Fifth University, Rabat, Morocco ' Laboratory LSE2I, National School of Applied Sciences (ENSAO), Mohammed First University, Oujda, Morocco ' Al-Qualsadi Research and Development Team, National Higher School for Computer Science and System Analysis (ENSIAS), Mohammed Fifth University, Rabat, Morocco
Abstract: The purpose of any effective supply chain is to find balance between supply and demand by coordinating all internal and external processes in order to ensure delivery of the right product, to the right customer, at the best time and with the optimal cost. Therefore, the estimation of future demand is one of the crucial tasks for any organisation of the supply chain system who has to make the correct decision in the appropriate time to enhance its commercial competitiveness. In an earlier study, where various artificial neural networks' structures are compared including perceptron, adaline, no-propagation, multi layer perceptron (MLP) and radial basis function for demand forecasting, the results indicate that the MLP structure present the best forecasts with the optimal error. Consequently, this paper focuses on realising a daily demand predicting system in a supermarket using MLP by adding inputs including previous demand, days' classification and average demand quantities.
Keywords: artificial intelligence; supply chain management; SCM; logistics; neural networks; demand forecasting; short-term forecasting.
International Journal of Logistics Systems and Management, 2017 Vol.28 No.2, pp.144 - 163
Received: 24 Mar 2016
Accepted: 05 Jun 2016
Published online: 04 Sep 2017 *