If they're basing this off browser user agents like a third of my traffic at least is random Chinese ips and requests with Chinese language in the headers.

All of them would be flagged as unknown by the major user agent parsing libraries.

None of the sure I work with have any Chinese content or content remotely useful for Chinese people, so it's scrapers.

Can it be fingerprint spoofers skewing the results? In that case the data is worthless.

That's probably Linux too

I think it's pretty clear from this graph that the macOS is eating their lunch, and to a lesser extent, Linux.

Android?

Hmm… What’s eating Windows’ lunch? Oh “Unknown” and “Other”. Cool.

If the data is coming from statcounter, it's kind of vague where they're pulling it.

It's strange to see IOS and not Android, maybe because it's fractured into so many sources

It could also be anti ad/tracker stuff.