The Semisupervised Approach for Data Driven and Consumer Oriented Sizing Systems in the Clothing Industry

In the fashion industry, garment sizing is a challenging issue and involves different problems resulting in lost profits. The design of a sizing system is a complex task involving both anthropometric data and subjective data, such as comfort or fit. With the emergence of big data, companies can collect valuable information about their consumers and, thus, enhance their sizing systems. Although several methods have been proposed in the literature, they do not integrate consumer comfort and fit. The method proposed in this article aims to design a suitable sizing system, while taking into account both morphological features and consumer preferences. Consumer preferences are extracted from a satisfaction survey, and are then integrated into an optimization process. The main contribution of the proposed model is twofold: a comprehensive evaluation of sizing systems that are based on consumer preferences is proposed and an optimization algorithm is developed to perfectly fit the problem. The proposed system is implemented and evaluated on the measurement database of the CAESAR project and a satisfaction survey conducted by a French fashion retailer. A comparison with up to date techniques and the existing sizing systems demonstrates that our model provides significant improvements for consumer fit and coverage.