Laver's Law Revisited: A Data-Driven Investigation of Fashion Trends

I'm not sure whether their ToS would allow it but Vogue's digitized archive[1] would be a great resource to scrape for photos and detailed descriptions of fashion trends over time. The archive goes all the way back to 1892 and allows you to filter the pages of each issue by article vs advertisement and do pretty comprehensive searches. For example, a search for "little black dress"[2] returns 1,969 images, 1,174 garments, 1462 ads, 11,531 articles, and only one cover from over a century ago.

It's a neat little slice of life regardless but I bet there's also a lot of cool data hiding in there.

[1]https://archive.vogue.com/issues/

[2]https://archive.vogue.com/search?QueryTerm=little%20black%20...

Pretty neat study. I like the humor in the trend descriptions. But it looks like pretty much every "trend" is a short-cycle sawtooth waveform which ends up telling me nothing at all. Maybe too many clusters, or they are different from the way designers think about it? Or does fashion really cycle so fast?