A Comparative Study of Machine Learning and Statistical Methods for Demand Forecasting in Supply Chains

Abstract

The selection of an appropriate demand forecasting method is crucial for effective supply chain management (SCM), as accurate forecasts help optimize inventory, production, and logistics. However, in markets characterized by high uncertainty and constant change, traditional statistical techniques, such as the ARIMA model, may not be sufficient to generate reliable and accurate forecasts. In response to this challenge, artificial intelligence (AI) algorithms, such as artificial neural networks (ANN) and random forests, offer promising solutions to improve forecasting accuracy. Despite this potential, the existing literature often provides only general descriptions of AI methods without comparing their performance in demand forecasting. This paper thus offers a comparative analysis of three main approaches used for demand forecasting : artificial neural networks, random forests, and the ARIMA model, evaluating their respective performances in the context of supply chain management. By assessing the strengths and limitations of each method, this study aims to provide valuable insights into their effectiveness and help companies choose the most suitable technique for their demand forecasting needs.