Demand Forecasting for Multichannel Fashion Retailers by Integrating Clustering and Machine Learning Algorithms
I-Fei Chen, Chi-Jie Lu
Abstract
In today’s rapidly changing and highly competitive industrial environment, a new and emerging business model—fast fashion—has started a revolution in the apparel industry. Due to the lack of historical data, constantly changing fashion trends, and product demand uncertainty, accurate demand forecasting is an important and challenging task in the fashion industry. This study integrates k-means clustering (KM), extreme learning machines (ELMs), and support vector regression (SVR) to construct cluster-based KM-ELM and KM-SVR models for demand forecasting in the fashion industry using empirical demand data of physical and virtual channels of a case company to examine the applicability of proposed forecasting models. The research results showed that both the KM-ELM and KM-SVR models are superior to the simple ELM and SVR models. They have higher prediction accuracy, indicating that the integration of clustering analysis can help improve predictions. In addition, the KM-ELM model produces satisfactory results when performing demand forecasting on retailers both with and without physical stores. Compared with other prediction models, it can be the most suitable demand forecasting method for the fashion industry.