Evaluation of deep learning with long short-term memory networks for time series forecasting in supply chain management

Abstract Performance analysis and forecasting the evolution of complex systems are two challenging tasks in manufacturing. Time series data from complex systems capture the dynamic behaviors of the underlying processes. However, non-linear and non-stationary dynamics pose a major challenge for accurate forecasting. To overcome statistical complexities through analyzing time series, we approach the problem with deep learning methods. In this paper, we mainly focus on the long short-term memory (LSTM) networks for demand forecasts in supply chain management, where the future demand for a certain product is the basis for the respective replenishment systems. This study contributes to the literature by conducting experiments on real data to investigate the potential of using LSTM networks for final customer demand forecasting, and hence for increasing the overall value generated by a supply chain. Both forward LSTM and bidirectional LSTM (forward-backward) for short- and long-term demand prediction in supply chain management are considered in this study