Proceedings of ICLT 2022
Demand Forecasting in Retail Based on Deep Learning Models
Panupong Wanjantuk; Temsiri Sapsaman; Ruth Banomyong; Prarawan Senachai
Department of Computer Engineering, Khon Kaen University, Thailand; Department of Production and Robotics Engineering, King Mongkut's University of Technology North Bangkok, Thailand; Department of International Business, Logistics and Transport, Thammasat University, Thailand; Khon Kaen University, Thailand
International Conference on Logistics & Transport 2022, Krabi, Thailand, pp. 165-169
Download PDF | View interactive page
Abstract
Forecasting future demand is essential to making supply chain decisions. Time series forecasting methods use historical demand to produce a forecast. The purpose of this paper is to apply a deep learning model called Long Short-Term Memory (LSTM) using historical sales data from a Thailand retailer for forecasting sales in retail supply chain. The data used in the analysis includes 46 months of actual daily sales of selected consumer goods. Based on this data, LSTM models have been trained and evaluated by using different look-back window sizes and different amounts of time to forecast the future sales. The result of the research is the appropriate deep learning models and parameters for demand forecasting that concludes from the analysis and evaluation of sales forecasting in the retail supply chain based on LSTM models. For each model, the relationships between the performance and the parameters, the look-back window sizes and the number of predicted time points into the future, are presented. Accurate demand forecasting in consumer goods could increase the competitive power of a retailer and improve its performance. The forecasting model can be used by businesses to optimize their inventory level, increase their bargaining power for purchasing, and ensure product availability. A novelty is the use of the LSTM model trained on real transaction data from a retail company that has based its business on the supply chain with suppliers and recipients in Thailand. perishable products, the blockage of storage space, and needle handling, including planning efforts [5].
Keywords
Demand Forecasting; Deep Learning; Supply Chain Management; Time Series; LSTM
Citation
Panupong Wanjantuk; Temsiri Sapsaman; Ruth Banomyong; Prarawan Senachai (2022). Demand Forecasting in Retail Based on Deep Learning Models. Proceedings of the International Conference on Logistics & Transport (ICLT 2022), Krabi, Thailand, pp. 165-169.