Yes you don't need to explain my field to me, thank you very much. Are you going to post wikipedia links next? It still doesn't change the fact that PCA is extremely unrigorous for any analysis that goes beyond 'looking at pretty figures'. The vectors are literally and utterly meaningless. They're selected in an ad hoc way for each dataset. If you want to untangle population structure properly there are countless other more rigorous methods, although they are not without their flaws. But the fact that you want to use PC1 and PC2 plots to prove a weird point about ethnicity makes me think you either don't know what you're talking about and are just using the field as a front to advance your weird beliefs. Even Steve Hsu would be more rigorous than that
>Do you have a reason to think that means PCA is bad at representing similarity and kinship because of this?
I suggest you read up on the definition of "distance" and "similarity" before drinking the PCA kool-aid. You don't get to define an ad hoc distance just because it fits your ideas about ethnicity. But then, I only have the popgen community to back me up on this. What do you have?
>As for the Slate Star Codex source - that wasn't to prove anything, just to define the "weak man" term.
SSC, providing ammunition to online HBD proponents since 2013.
Now my opinion may seem controversial but I personally believe having feature requests that are old enough to legally drink is not a sign of healthy project management
Pretty moribund. They got dropped by Mozilla so it's entirely community-driven, and there's only so many things you need to add to an email client, unless you want it to turn into emacs or something.
That said it's a damn good product that's damn good at what it does.
Just in case people don't know, although the name sounds French I've literally never seen one in France. I guess it's like Mexico's Taco Bell or something
Using a PCA to prove an intrinsic point about ethnicity is as laughable as it gets. You're aware the vectors in a PCA are just linear combinations of the observed features, right? You're just staring at the same data a different way.
But of course I didn't expect any less from someone who quotes Slate Star Codex of all sources to prove a point.
If you've just now taken the venv plunge wait until you learn about conda. Python is more or less a shitshow when it comes for packaging and it's the reason I'm drifting away from it little by little
The actual solution to gerrymandering is to discard the whole silly idea of 'district' and vote proportionally, just many countries do and are not the worse off for it.
What about the rural states, you say? Well they still get to manage their own, internal affairs with their own legislature, governors, etc. But as far as state-level decisions are concerned, their weight should be exactly equal to that of their population representation.
The reason France's math output is so stellar is because of the controversial grand école system - pretty much every great French mathematician you can name came out of one of the top 2 schools. It's very elitist (and the elitism is misplaced since the system as well as the schools are virtually unknown outside of France compared to, say, the Ivies) but it does have results.