https://i.postimg.cc/NFH8ZSQW/image.png
https://i.postimg.cc/HxyDPW7H/image.png
Printable View
could you make simplified Bulgarian regional averages?
maybe along these lines:
https://upload.wikimedia.org/wikiped...r_Bozhinov.png
I agree with this approach, within one nation there should be differentiations based on regions (8-10 at max) rather than counties or cities (with the exception of the capital, which is often the meeting point of all regions), otherwise we get way too many results from the same nation.
I believe 5 samples should be absolute minimum to create any separate cluster, and even that is kinda on the borderline. Now I do wonder if anyone still even knows on how many samples are the various groupings even created on? Rather have one or two well representative national categories than 7-8 based on 2-3 samples each.
In my opinion 3 samples should be absolute minimum for one average, around 10 is decent. 1 is unacceptable, that person can be atypical for the region and that's not average.
Do you have the PCA with the new french sample?