Elimination of outliers in GEDmatch calculators’ database?

[reboun]

As far as I know, when forming the database for GEDmatch calculators, samples are collected for each ethnic group. When sufficient number of samples are collected, the outliers are eliminated. My question is, how accurate is to eliminate the outliers. Think of an ethnic group which is genetically very diverse. Say, 1000 samples are collected among the ethnic group but bevause of the genetic diversity, a lot of samples are classified as outlier and the database is left with only 150 samples. Wouldn’t it be a loss of accuracy?

In my opinion, outliers should also be included in GEDmatch calculators’ database.

[princeton90]

Such diversity can only be found in New World populations and the New World Populations’ results does not exist in any of the GEDmatch calculators.

[JamesBond007]

As far as I know, when forming the database for GEDmatch calculators, samples are collected for each ethnic group. When sufficient number of samples are collected, the outliers are eliminated. My question is, how accurate is to eliminate the outliers. Think of an ethnic group which is genetically very diverse. Say, 1000 samples are collected among the ethnic group but bevause of the genetic diversity, a lot of samples are classified as outlier and the database is left with only 150 samples. Wouldn’t it be a loss of accuracy?

In my opinion, outliers should also be included in GEDmatch calculators’ database.

GEDmatch is outdated and even though G25 is not quite 'cutting edge' , anymore, it is still the best tool available to the layman. One can remove or add outliers, obviously, unlike GEDmatch. Interestingly enough two of the best so called calculators on GEDmatch are Dodecad K12B and Eurogenes K13 the former says I'm 'mixed Germanic' and the latter says I'm closest to the Irish -- that is the kind of retarded bullshit that happens with GEDmatch. K36 still has a role when finer granularity is needed in a minority of use cases e.g. can't distinquish Finnish-mongol genes from East German genes so on and so forth but in general GEDmatch is outdated crap.

BTW, just or the record my closest G25 group is Dutch or Central_Dutch and that is more inline with my K12B results and K36 Tolan map results and while I'm not Dutch there are no Dutch or English genes just ADMIXTURE(s) that are most typical of those populations hence some English people will skew more towards Danish or Dutch.

More to the point it makes more sense to compare yourself to generalized averages than to outliers in general I think although G25 offers you the flexibility to do the latter.

[Ion Basescul]

Such diversity can only be found in New World populations and the New World Populations’ results does not exist in any of the GEDmatch calculators.

The Moldovan average has full Ukrainians in it.

[reboun]

The Moldovan average has full Ukrainians in it.

How is it known that they are fully Ukrainian?