Machine learning for demand forecasting in the physical internet: a case study of agricultural products in Thailand

Abstract

Supply chains are complex, stochastic systems. Nowadays, logistics managers face two main problems: increasingly diverse and variable customer demand that is difficult to predict. Classical forecasting methods implemented in many business units have limitations with the fluctuating demand and the complexity of fully connected supply chains. Machine Learning methods have been proposed to improve prediction. In this paper, a Long Short-Term Memory (LSTM) is proposed for demand forecasting in a physical internet supply chain network. A hybrid genetic algorithm and scatter search are proposed to automate tuning of the LSTM hyperparameters. To assess the performance of the proposed method, a real-case study on agricultural products in a supply chain in Thailand was considered. Accuracy and coefficient of determination were the key performance indicators used to compare the performance of the proposed method with other supervised learnings: ARIMAX, Support Vector Regression, and Multiple Linear Regression. The results prove the better forecasting efficiency of the LSTM method with continuous fluctuating demand, whereas the others offer greater performance with less varied demand. The performance of hybrid metaheuristics is higher than with trial-and-error. Finally, the results of forecasting model are effective in transportation and holding costs in the distribution process of the Physical Internet.

KEYWORDS:

• Demand forecasting
• machine learning
• recurrent neural network
• physical internet
• distribution
• metaheuristic

Acknowledgements

Regarding the successful results in this research work, we would like to thank an internship student, Niama Boumzebra, for preparing the dataset to test the forecasting performance with regression methods. We would also like to thank the Office of Agricultural Economics Thailand for providing the initial dataset for generating data in this experiment. Finally, we would like to thank Campus France and Burapha University for their sponsorship.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Anirut Kantasa-ard

Anirut Kantasa-ard received his Bachelor’s degree in computer science (2010) from King Mongkut’s University of Technology Thonburi (Thailand), Master degree in management and Optimisation in Supply Chain (2016) and Transport from IMT Atlantique, Nantes (France). Anirut Kantasa-ard is currently a Ph.D. student at the Polytechnic University of Hauts-de-France. His research interests include Machine Learning and implementation the optimisation for logistics and supply chain management. He is the author of 3 publications in international conferences in the field of production and supply chain management.

Maroua Nouiri

Maroua Nouiri received her Bachelor’s degree in computer science in (2010) from Faculty of sciences of Gabes, Master degree in Computer Science and Multimedia from Higher Institute of Computer Science and Multimedia of Gabes (2012), and her Ph.D. (2017) in Electronics, Information and Communication Technologies from Polytechnic School of Tunisia. She worked as a research engineer (Post-doctoral) on sustainable development in logistics systems. Maroua Nouiri is currently an Associate Professor at the University of Nantes, France. Her research areas concern the control and the optimisation of discrete event systems (manufacturing, logistics, transport) using metaheuristic, multi-agent systems and AI tools. She is the author of more than 5 articles in international journals and 13 peer reviewed publications in chapters of books and international conferences in the field of intelligent manufacturing and distribution systems and she is involved in several research projects.

Abdelghani Bekrar

Abdelghani Bekrar received his engineering degree in computer science (1999) from INI (Algerian National high school), Master degree from ECN (French National high school), and his Ph.D. (2007) from the University of Technology of Troyes (UTT), France. Abdelghani Bekrar is currently an Associate Professor at the Polytechnic University of Hauts-de-France. His research interests include Metaheuristic design and implementation and Hard optimisation for engineer applications and supply chain management. He is the author of more than 32 articles in international journals and 45 publications in international conferences in the field of manufacturing and transportation systems and he is involved in several research projects.

Abdessamad Ait el cadi

Abdessamad AitElCadi received his engineering degree in ENPC from Ponts et Chaussées National high school, France (2000), Master degree from Applied Science (École Polytechnique de Montréal, Canada), and his Ph.D. (2010) from École Polytechnique de Montréal, Canada. Abdessamad AitElCadi is currently an Associate Professor at the Polytechnic University of Hauts-de-France. His research interests include modelling, simulation, and optimisation in decision-making process in industry. He is the author of more than 10 articles in international journals and 17 publications in international conferences in the field of manufacturing and transportation systems and he is involved in several research projects.

Yves Sallez

Yves Sallez is currently Professor in INSA Engineering School of the Polytechnic University of Hauts-de-France, where he teaches robotics, vision and automated production. He received his PhD in automatics from the Valenciennes University in 1988 and his HDR (French diploma authorising supervision of PhD student and enabling access to full professor status) in 2012. He is member of the LAMIH at the Polytechnic University of Hauts-de-France, partner of a French Carnot network (ARTS). He is also a member of the French CNRS research group, MACS. He is the author of more than 22 articles in international journals and 60 publications in international conferences in the field of manufacturing and transportation systems. His research interests are in the area of intelligent/active products and heterarchical systems with applications in manufacturing, transport and logistics.