Quote:
Originally Posted by Chris596
I see no sufficient reason to ignore or invalidate the results with unofficial ("simulated") coordinates.Quote:
Originally Posted by gixajo
First of all: I think it's correct and commendable for some to pay Davidski for the official coordinates.
I didn't do that; my country's currency is terrible, and I would have to pay for currency conversion and transfer fees to the bankers. I'm sorry, but since the "simulated coordinates" work well for pre-selected models (discarding improbable samples), I'm satisfied with what I have.
Furthermore, I opened a thread asking forum members to show results from different coordinates of their own (official, simulated, K13, K15, etc.), but I think I wasn't clear, as nobody understood (my English is not so good) or wanted to participate in my test correctly.
https://www.theapricity.com/forum/sh...ht=coordinates
Some questions:
Is the official one really more accurate than the simulated ones?
If so, which simulated ones?
What does it mean to be accurate?
Is it detecting more populations or fewer?
If the simulated data is less accurate, does that mean it produces false results every time?
By making the necessary adjustments to the calculator model, is it possible to obtain reliable results?
We know that "raw data" is more sensitive, and this has a smaller distance and detects more populations in large models.
The "scaled" data already has an artificial distance to avoid noise.
If the simulated data is worse because it detects more, one has the option to manually increase the distance in Vahaduo.
In my last participations in this forum, I have used my simulated data from different sources to show that they are consistent, although K13 and K15 detect less than the K36 simulated data and the MyHeritage simulated data. And the distances of the MH data are usually smaller, but this can be solved by artificially adding distance, right?
Finally, with the appropriate models, I get results very similar to my MH v0.95 results, which are atypical for Brazilians from the region where I was born. And the results are supported by my family history, so my "simulated" G25 coordinates give me predicted results.
At least regarding the question: By making the necessary adjustments to the calculator model, is it possible to obtain reliable results?
The answer seems to be "yes."
If my result were random, for example, showing a high percentage of Japanese or Indian, then it would be a lottery. But the imprecision is not absurd, sometimes linking me to the populations of southeastern Europe, with whom I share genetic similarities due to my Mediterranean and Slavic ancestry.
And it is known that even those with official coordinates need to make such adjustments.
