AI based forecasting in fast fashion industry: a review

Abstract

Abstract The textile and clothing sector is considered as one of the driving sectors of the Moroccan economy. It occupies a highly strategic position given the importance of its contribution in terms of growth and jobs. This sector, which is undergoing profound changes essentially linked to progressive dismantling of the Multi-Fibres arrangements (MFAs) that have governed trade long years, remains by far the largest export sector in the Moroccan industry. In this environment where competition will become more and more fierce, the ideal for a textile company is to produce exactly the products that its customers will buy. This ideal is unfortunately impossible, except in very special case where the company starts to manufacture from the receipt of the order of the client. Thus, in order to make the decisions relating to its proper functioning and to its durability, the company must rely on a reliable forecasting system. Such a system allows to achieve a competitive advantage by minimizing supply chain costs, avoiding shortages and minimizing financial losses due to the balance of unsold items. The purpose of this article is to make a review of the literature on fashion retail sales forecasting methods. We have explored different kinds of analytical methods for fast fashion retail sales forecasting, their features and characteristics. From the reviewed literature we got a synthetic view of the pure statistical models, the pure Artificial Intelligence (AI) models, and the hybrid models.