Demand Forecasting in Supply Chain Management using CNN-LSTM Hybrid Model

Abstract

Efficient supply chain management is necessary to meet customer demands. Demand forecasting is a predictive analysis that estimates how much of a product or service a customer will need in the future. Accurate demand forecasting relies on past sales data and statistical estimates from test markets. Deep neural networks have gained popularity as a demand forecasting technique in supply chain management because of their high level of accuracy. The objective of this study is to use Long Short-Term Memory (LSTM) neural networks with hyperparameter tuning, as well as a hybrid model comprising Long Short-Term Memory and 1D Convolutional Neural Networks (CNNs), to predict the demand for eight medicinal drugs. The results obtained from the hybrid model will be compared with the LSTM models to select the best model for long-term forecasting. Demand forecasting has several applications, including inventory control, production planning, and market entry decisions