MTA: PCA Plot (Can someone make sense out my father's plot)

[Kyp]

https://i.imgur.com/HOxTPvQ.jpg

This seems a bit strange


1. Azeri (7.627)
2. Turk (7.641)
3. Kurdish (8.250)
4. Iranian (8.648)
5. Turkish (11.10)
6. Greek_Azov (12.32)
7. Cirkassian (13.14)
8. Kumyk (13.61)

[Lucas]

https://i.imgur.com/HOxTPvQ.jpg

This seems a bit strange

He plots with Azeris generally ? What is his k36?

[Kyp]

He plots with Azeris generally ? What is his k36?

Yes, even the numeral distances on MTA have Azeri as number1 but with pretty high distance (Iranians are closer to the reference for example), Same for Dodecad

K36:
[1] "1. CLOSEST SINGLE ITEM DISTANCE%"
AZE_Azerbaijan
0.8205413
TUR_Kurdistan
0.8495381
IR_West-Azerbaijan
0.8807537

[Hapanuwa]

Don't bother searching for logic in MTA PCA plots, brother. Their PCA plots seem very strange and off to me as well.

My results are weird too
https://i.postimg.cc/wx1ZG4YB/Screenshot-20200511-000845-Gallery.jpg (https://postimg.cc/SjpgXrs0)

[Kyp]

Don't bother searching for logic in MTA PCA plots, brother. Their PCA plots seem very strange and off to me as well.

My results are weird too


I atleast can make sense out of yours. Between Greek_Crete and Turkmen

[Hapanuwa]

I atleast can make sense out of yours. Between Greek_Crete and Turkmen

True. I couldn't explain your father's plot though even if I would try lol

[Kaspias]

The most sensible option you can do is closing MTA tab

[Halgurd]

Their PCA plot is wrong on so many levels. On mine it says Azeris cluster with Georgian Jews.

Edit: i just saw it says Georgian Laz not jews

[Kyp]

Their PCA plot is wrong on so many levels. On mine it says Azeris cluster with Georgian Jews.

Can you post yours for comparison? I'm trying to understand them better. They have to follow some kind of logic...

[Halgurd]

Can you post yours for comparison? I'm trying to understand them better. They have to follow some kind of logic...

https://i.ibb.co/jZrfy1z/AC2-DE479-FF5-A-4-DCD-989-C-6-AE6-ED4-DE397.jpg

[Zoro]

https://i.imgur.com/HOxTPvQ.jpg

This seems a bit strange


1. Azeri (7.627)
2. Turk (7.641)
3. Kurdish (8.250)
4. Iranian (8.648)
5. Turkish (11.10)
6. Greek_Azov (12.32)
7. Cirkassian (13.14)
8. Kumyk (13.61)

There is a couple of things you should know about PCAs:

First, you are only looking at 2 axis PC1 and PC2. You can't see where your father plots on the other axes compared to Azeris. PC3,PC4,PC5 etc. Even if you do how do you mentally add up all those effects and come up with a final location

Second, anyone who has run PCAs will tell you that every PCA will come out different depending on what samples you include in the run because other samples push and pull on your dad in different ways and it's impossible to predict or quantify what those pushes and pulls will be. there is a corect way of running PCAs but no one seems to follow it or they just don't care.

Third, this PCA seems extra weird because you can't even tell which direction is the E. Asian axis of variation and which one the W. Asian and S. Asian. I thought the left was E. Asian axis but then why are Turks and Azeris on the bottom of the PCA? And which direction is S. Asian variation ? looks weird.


At least with other PCAs like the one I received from Dilawer below a few months ago I can tell that SW is E. Asian axis and ESE is ENF axis but theirs is totally confusing

https://i.imgur.com/Vs6vkXC.jpg

[Kyp]

There is a couple of things you should know about PCAs:

First, you are only looking at 2 axis PC1 and PC2. You can't see where your father plots on the other axes compared to Azeris. PC3,PC4,PC5 etc. Even if you do how do you mentally add up all those effects and come up with a final location

Second, anyone who has run PCAs will tell you that every PCA will come out different depending on what samples you include in the run because other samples push and pull on your dad in different ways and it's impossible to predict or quantify what those pushes and pulls will be. there is a corect way of running PCAs but no one seems to follow it or they just don't care.

Third, this PCA seems extra weird because you can't even tell which direction is the E. Asian axis of variation and which one the W. Asian and S. Asian. I thought the left was E. Asian axis but then why are Turks and Azeris on the bottom of the PCA? And which direction is S. Asian variation ? looks weird.


At least with other PCAs like the one I received from Dilawer below a few months ago I can tell that SW is E. Asian axis and ESE is ENF axis but theirs is totally confusing



I understand all that. Still trying to find some logic though

[Kyp]

There is a couple of things you should know about PCAs:

First, you are only looking at 2 axis PC1 and PC2. You can't see where your father plots on the other axes compared to Azeris. PC3,PC4,PC5 etc. Even if you do how do you mentally add up all those effects and come up with a final location

Second, anyone who has run PCAs will tell you that every PCA will come out different depending on what samples you include in the run because other samples push and pull on your dad in different ways and it's impossible to predict or quantify what those pushes and pulls will be. there is a corect way of running PCAs but no one seems to follow it or they just don't care.

Third, this PCA seems extra weird because you can't even tell which direction is the E. Asian axis of variation and which one the W. Asian and S. Asian. I thought the left was E. Asian axis but then why are Turks and Azeris on the bottom of the PCA? And which direction is S. Asian variation ? looks weird.


At least with other PCAs like the one I received from Dilawer below a few months ago I can tell that SW is E. Asian axis and ESE is ENF axis but theirs is totally confusing



I understand all that. Still trying to find some logic though