Visualisation for analysing evolutionary dynamics of fashion trends

This study explores how to analyse evolutionary dynamics of fashion trends, focusing on trickle-down process from luxury fashion products. To reveal chronological changes of fashion trends, this paper proposes visualisation techniques using networks created by quantitative analysis of fashion trends. For quantitative analysis, 165 fashion trends comprised of 495 pictures were collected from articles of online magazine Vogue Paris summarising major fashion weeks from 2013 S/S to 2016 A/W. Two types of network visualisation techniques were discussed, which utilised cluster analysis as a hierarchical clustering technique and k-means as a non-hierarchical clustering technique. A comparison between two clustering techniques showed a drawback of hierarchical clustering caused by cluster integration order. The networks created by the non-hierarchical clustering technique unveiled both steady and transient trends. It was confirmed that the proposed method provided an understanding of the evolutionary dynamics of fashion trends.