The use of AI for demand and trend forecasting in fashion and the potential of these methods for sustainable brands

Abstract

Sustainable fashion brands have faced struggles in setting up supply chains and responding to trends, while consumers constantly prioritize other attributes in clothing than sustainability. This thesis discusses the possibility of sustainable brands using AI forecasting tools to address the common issues sustainable brands face. In this thesis, the fit of an AI method for sustainable brands is evaluated by how useful it could be in design decision making regarding conventional clothing features that include price or price-performance ratio, quality, texture, design, colour, fit and comfort. AI forecasting has been widely researched in the context of fashion, and specifically fast fashion, but its possibilities for sustainable fashion have not been widely researched. The research question of this thesis is “How is AI used in fashion forecasting and what potential does AI have for sustainable brands in terms of fashion forecasting?”. Sub-questions of this thesis are “What AI methods are commonly used in fashion forecasting, and how could each method fit the needs of sustainable brands?” and “What needs could sustainable brands have in terms of forecasting?”. The aim of this thesis is to provide an overview of AI-based methods used in trend and demand forecasting in the fashion industry and discuss whether AI-based methods have potential for sustainable fashion brands. This thesis discusses common AI methods reviewed or researched in the context of fashion, which include artificial neural networks (ANN), evolutionary neural networks (ENN), convolutional neural networks (CNN), extreme learning machine (ELM) and fuzzy logic (FL). In addition, hybrid methods in the context of fashion forecasting are introduced. In addition to that, AI-methods are evaluated by their ability to produce long-term trend forecasts to help creating timeless designs for durable clothing. The AI methods introduced in this thesis could fit the needs of sustainable brands, but additional research is needed to study how AI methods fit sustainable brands in practice.