G25 ancient model for Stears

[Jana]

His lowest distance ever (EXTREMELY LOW, ALMOST PERFECT FIT!), took me a lot of time to figure this out :)

Target: Stears
Distance: 0.1850% / 0.00185012

41.8 Balto-Slavic
24.8 Old_Balkan
13.6 Near_Eastern
11.6 Uralic
3.6 Germanic
2.4 North_Caucasian
2.2 Mongoloid

[gixajo]

His lowest distance ever (EXTREMELY LOW, ALMOST PERFECT FIT!), took me a lot of time to figure this out :)

Target: Stears
Distance: 0.1850% / 0.00185012



Very very close distance,for him:thumb001:

Could you post the sources so anyone can try it?

[Narration]

Which software did you use?

[Jana]

Very very close distance,for him:thumb001:

Could you post the sources so anyone can try it?

It's very complicated, and each model takes a lot of time to make, because I use full gradient run (entire individual ancient samples spreadsheet). It's very individualised, can't be used as generic run :p

[Jana]

Which software did you use?

Vahaduo custom G25 Calculators, ancients spreadsheet (individual samples not averages)

[Ion Basescul]

"Overfitting is a modeling error that occurs when a function is too closely fit to a limited set of data points. Overfitting the model generally takes the form of making an overly complex model to explain idiosyncrasies in the data under study."

You should aim for a distance of 2 ideally.

[Jana]

"Overfitting is a modeling error that occurs when a function is too closely fit to a limited set of data points. Overfitting the model generally takes the form of making an overly complex model to explain idiosyncrasies in the data under study."

You should aim for a distance of 2 ideally.

Meh, changed opinion on that after XP made some excellent points. It's not overfitting, it's entire spreadsheet run.

[Ion Basescul]

Meh, changed opinion on that after XP made some excellent points. It's not overfitting, it's entire spreadsheet run.

I have no idea who XP and his excellent points are. But you can read extensively about overfitting from Generallisimo/Davidski (creator of Global 25) and Huijbregts (creator of nMonte, which is the algorithm that runs the coordinates on Vahaduo/R).
Just google it.

[Defcon2]

His lowest distance ever (EXTREMELY LOW, ALMOST PERFECT FIT!), took me a lot of time to figure this out :)

Target: Stears
Distance: 0.1850% / 0.00185012

41.8 Balto-Slavic
24.8 Old_Balkan
13.6 Near_Eastern
11.6 Uralic
3.6 Germanic
2.4 North_Caucasian
2.2 Mongoloid

Scaled? I would also like to know the most fit model for me.

[Jana]

I have no idea who XP and his excellent points are. But you can read extensively about overfitting from Generallisimo/Davidski (creator of Global 25) and Huijbregts (creator of nMonte, which is the algorithm that runs the coordinates on Vahaduo/R).
Just google it.

I ran entire spreadsheet without removing any samples, and that's why this model is so tight. I don't believe in any models which handpick samples they like and force algorithm to chose them.
He won't get any better model than mine :)

[Jana]

Scaled? I would also like to know the most fit model for me.

Yeah. It's very hard job, you need to run entire spreadsheet (all ancients available on G25) and than assign what you get to culture that makes sense for it. 0 penaliation btw.
For example, if he scores Avar_Szolad I would move it to Balto-Slavic, if he scored Bulgarian_IA I classed it as Old_Balkan etc.

Very delicate and hard job. I did the same for me, my results are in my sig. Before that I always had horrible fits and could never get any close distances.

[Defcon2]

Yeah. It's very hard job, you need to run entire spreadsheet (all ancients available on G25) and than assign what you get to culture that makes sense for it.
For example, if he scores Avar_Szolad I would move it to Balto-Slavic, if he scored Bulgarian_IA I classed it as Old_Balkan etc.

Very delicate and hard job. I did the same for me, my results are in my sig. Before that I always had horrible fits and could never get any close distances.

I like that idea :)

[Jana]

I like that idea :)

Also if you get samples which you don't know/aren't sure what they are, you must run them compared to moderns, so you will be able to identify the culture by the fact to which modern pops they are closest (both distance and modeling)

[gixajo]

Vahaduo custom G25 Calculators, ancients spreadsheet (individual samples not averages)

In unscaled G25 Ancient Individual Samples, official datasheet, I get a very close distance too, but with a longer list of results, without "customize" it.

Target: gixajo
Distance: 0.2463% / 0.00246336
15.6 Bell_Beaker_NLD:I4068
12.8 Iberia_Northeast_c.6-8CE_ES:I7673
11.6 Iberia_Southeast_MLN:I7600
8.2 DEU_MA:AED_106
5.4 FRA_BA:QUIN58
5.4 Iberia_Southeast_MLN:I7594
5.0 Iberia_Northeast_c.6CE_PL:I12034
4.8 HUN_Starcevo_N:I1876
3.6 Iberia_Menorca_LBA:I3315
3.0 DEU_LBK_N:I2008
2.8 RUS_Saltovo-Mayaki_low_res:DA190
2.6 DEU_MA:STR_316
2.6 England_CA_EBA:I5373
2.6 ITA_Sardinia_Nuragic:I10365
1.8 FRA_IA:NOR3-15
1.8 Iberia_Northeast_RomP:I8339
1.6 Canary_Islands_Guanche:gun005
1.4 MAR_Taforalt:TAF010
1.4 RUS_Afanasievo:I6711
1.4 SWE_Viking_Age_Sigtuna:vik_gtm127
1.2 Canary_Islands_Guanche:gun008
1.2 CHE_FN_steppe:MX304
0.8 ITA_Collegno_MA:CL121
0.6 ZAF_2100BP:I9028
0.4 PAK_Katelai_IA:I12472
0.2 Bell_Beaker_CHE:I5759
0.2 Levant_ISR_Ashkelon_LBA:ASH33

[Ion Basescul]

I ran entire spreadsheet without removing any samples, and that's why this model is so tight. I don't believe in any models which handpick samples they like and force algorithm to chose them.
He won't get any better model than mine :)

It's only natural to get a very close fit with a lot of samples. But that doesn't indicate that some of those samples who result in the overfit contributed any real ancestry. For example in your case, it's very unlikely that you have any real North Caucasian ancestry. That's probably limited only to bordering regions in Russia and Ukraine.

[Jana]

In unscaled G25 Ancient Individual Samples, official datasheet, I get a very close distance too, but with a longer list of results, without "customize" it.

Target: gixajo
Distance: 0.2463% / 0.00246336
15.6 Bell_Beaker_NLD:I4068
12.8 Iberia_Northeast_c.6-8CE_ES:I7673
11.6 Iberia_Southeast_MLN:I7600
8.2 DEU_MA:AED_106
5.4 FRA_BA:QUIN58
5.4 Iberia_Southeast_MLN:I7594
5.0 Iberia_Northeast_c.6CE_PL:I12034
4.8 HUN_Starcevo_N:I1876
3.6 Iberia_Menorca_LBA:I3315
3.0 DEU_LBK_N:I2008
2.8 RUS_Saltovo-Mayaki_low_res:DA190
2.6 DEU_MA:STR_316
2.6 England_CA_EBA:I5373
2.6 ITA_Sardinia_Nuragic:I10365
1.8 FRA_IA:NOR3-15
1.8 Iberia_Northeast_RomP:I8339
1.6 Canary_Islands_Guanche:gun005
1.4 MAR_Taforalt:TAF010
1.4 RUS_Afanasievo:I6711
1.4 SWE_Viking_Age_Sigtuna:vik_gtm127
1.2 Canary_Islands_Guanche:gun008
1.2 CHE_FN_steppe:MX304
0.8 ITA_Collegno_MA:CL121
0.6 ZAF_2100BP:I9028
0.4 PAK_Katelai_IA:I12472
0.2 Bell_Beaker_CHE:I5759
0.2 Levant_ISR_Ashkelon_LBA:ASH33


This is it. He and me got the same. Now you need to assign names to every sample based on logic and how they plot genetically, and than add all that togheder.
For example all old Iberian you give the same name and merge them togheder, all Germanic like too, all MENA like also etc.

Use your intuition and history/genetics knowledge.

[Jana]

It's only natural to get a very close fit with a lot of samples. But that doesn't indicate that some of those samples who result in the overfit contributed any real ancestry. For example in your case, it's very unlikely that you have any real North Caucasian ancestry. That's probably limited only to bordering regions in Russia and Ukraine.

And you know that how? My grandmother was not Croatian and she had very mixed and complex ancestry that I never learned completely about.
Since her ancestors lived in Budapest before moving to Vojvodina, they could have been mixed with anything, all kinds of people lived in that city.

I have high Kavkaz affinity on every calculator, so yeah. If I remove Kavkaz part my fits worsens considerably.

[17571imre]

This is it. He and me got the same. Now you need to assign names to every sample based on logic and how they plot genetically, and than add all that togheder.
For example all old Iberian you give the same name and merge them togheder, all Germanic like too, all MENA like also etc.

Use your intuition and history/genetics knowledge.

but what if the sample is mixed? and it has a mix of Germanic and slavic? where do you put it then

[Jana]

but what if the sample is mixed? and it has a mix of Germanic and slavic? where do you put it then

I didn't have such problems on 0 pen, it usually gives clean results compared to 0.25X (for example it gives me very med and very baltic samples with no in-betweeners)
If it's mixed you can give it modern name.

I actually had one such case, which was Germanic/Celtic mix and that's why I named it German, not Germanic. It wasn't clean but very modern south German like.
While Stears scored pure Germanic samples, thus his NW part is called Germanic and not German.

Hop you catch the drift :p

[gixajo]

This is it. He and me got the same. Now you need to assign names to every sample based on logic and how they plot genetically, and than add all that togheder.
For example all old Iberian you give the same name and merge them togheder, all Germanic like too, all MENA like also etc.

Use your intuition and history/genetics knowledge.

I read about a similar method in Anthrogenica.

So now I must group my list of results under a common denomination with some historic sense (for example all Iberian peninsula results as Celtiberian or anything similar, or Guanche and Mar_Taforalt together under another denomination), and put ":" , so that they add up and give higher percentages in results.

Thanks.



Thanks.

[Defcon2]

It's only natural to get a very close fit with a lot of samples. But that doesn't indicate that some of those samples who result in the overfit contributed any real ancestry. For example in your case, it's very unlikely that you have any real North Caucasian ancestry. That's probably limited only to bordering regions in Russia and Ukraine.

This tool I do not think is useful to measure the real ancestry but its similarities in the components. For example, I usually score with the Northern Italians (Etruscans) and Illyrians but I doubt that I have real ancestry of any of them.

[gixajo]

I didn't have such problems on 0 pen, it usually gives clean results compared to 0.25X (for example it gives me very med and very baltic samples with no in-betweeners)
If it's mixed you can give it modern name.

I actually had one such case, which was Germanic/Celtic mix and that's why I named it German, not Germanic. It wasn't clean but very modern south German like.
While Stears scored pure Germanic samples, thus his NW part is called Germanic and not German.

Hop you catch the drift :p

To clarify it, I say you thanks, because with your example and your little explanation with only 2 lines, it is easier to understand that method than a long technical explanation as I see in another internet sites.

[Jana]

This tool I do not think is useful to measure the real ancestry but its similarities in the components. For example, I usually score with the Northern Italians (Etruscans) and Illyrians but I doubt that I have real ancestry of any of them.

Yes, because those peoples were very similar to Iberians. So you can freely count these samples to your native Iberian part.
It's different in my case since north caucasian ancestry is completely different to anything south slavic or european like, these is no genetic similarity.

So I obviously have some ancestry from Caucasus. How modern or ancient it is, that I can't tell.

[Jana]

To clarify it, I say you thanks, because with your example and your little explanation with only 2 lines, it is easier to understand that method than a long technical explanation as I see in another internet sites.

You're welcome. I had no clue some other people got the same idea as me :)

[17571imre]

deleted

[Defcon2]

MLN what does it mean? there should be a guide on codes and names of these samples, this looks like the encryption of Matrix.

[Ion Basescul]

And you know that how? My grandmother was not Croatian and she had very mixed and complex ancestry that I never learned completely about.
Since her ancestors lived in Budapest before moving to Vojvodina, they could have been mixed with anything, all kinds of people lived in that city.

I have high Kavkaz affinity on every calculator, so yeah. If I remove Kavkaz part my fits worsens considerably.

It's simply very unlikely. Your Caucasus affinity should come from an elevated CHG or Iran_N and I am sure that it falls within the spectrum for your country, even if on average you might have more of it.

[Jana]

MLN what does it mean? there should be a guide on codes and names of these samples, this looks like the encryption of Matrix.

Probably middle late neolithic sample, run it with moderns to see how they plot and than it will be clearer.

[Jana]

It's simply very unlikely. Your Caucasus affinity should come from an elevated CHG or Iran_N and I am sure that it falls within the spectrum for your country, even if on average you might have more of it.

It's not from my Croatian part but from my mother who also scores very high CHG/west asian and she is 50% non Croatian.
My father who is pure Croat scores normally, much less than we do.

My mother's ancestry is very weird mixture of something I wasn't menage to identify yet properly.
Anyway it doesn't matter, I have Caucasus affinity and it it's there. Is it ancient or recent isn't particulary important to me.

[Ion Basescul]

It's not from my Croatian part but from my mother who also scores very high CHG/west asian and she is 50% non Croatian.
My father who is pure Croat scores normally, much less than we do.

My mother's ancestry is very weird mixture of something I wasn't menage to identify yet properly.
Anyway it doesn't matter, I have Caucasus affinity and it it's there. Is it ancient or recent isn't particulary important to me.

Try these ancient components model for you and your mom. You don't need to post the results here.
If that CHG is not contained within Yamnaya, then maybe there is something interesting that happened.
I know for example that my father scores more Caucasus-related ancestry than the rest of us, but his is contained within the steppe component.
Hence, it is unlikely that it was mediated via a source from Caucasus. Maybe that's the same in your case.



Anatolia_Barcin_N,0.1175998,0.180118,0.0035312,-0.101158,0.0510443,-0.0483875,-0.0043582,-0.0069334,0.0362287,0.0807473,0.0079718,0.0118803,-0.0234545,0.0004691,-0.0419807,-0.0101913,0.0233091,0.0019866,0.0136954,-0.0097489,-0.0142249,0.0057723,-0.0041232,-0.0031658,-0.0043437
GEO_CHG,0.091058,0.102568,-0.083344,-0.00323,-0.08617,0.020638,0.024911,-0.001846,-0.128236,-0.074717,-0.006333,0.023979,-0.054856,0.004404,0.026601,-0.03275,0.02386,-0.013429,-0.022249,0.034767,0.033815,-0.007048,0.006532,-0.025787,-0.002036
WHG,0.1246365,0.116278,0.184789,0.189279,0.1546445 ,0.0464355,0.0131605,0.0372675,0.0890705,0.017768,-0.0153455,-0.015811,0.0159065,-0.0030275,0.053338,0.0582065,0.00502,0.016343,-0.0093015,0.055589,0.0944585,0.0111905,-0.049607,-0.160866,0.0170045
Baltic_LVA_HG,0.1292603,0.1004104,0.1802636,0.1972 329,0.1064236,0.0560919,0.0083574,0.0207106,0.0524 476,-0.0261394,-0.0046788,-0.0181151,0.0281434,-0.0053759,0.0355759,0.0480141,0.003072,0.0040856,-0.0059236,0.042325,0.0613526,0.0130531,-0.0262979,-0.1110549,0.0095949
IRN_Ganj_Dareh_N,0.0430252,0.0664158,-0.1550722,0.0047158,-0.122669,0.0235384,0.017109,-0.0011998,-0.082546,-0.0544158,-0.0028258,-0.0016186,0.0044896,-0.0062756,0.0316498,0.0561384,-0.0054242,0.0068664,0.0136508,-0.0334162,0.00856,-0.028836,-0.0110678,-0.039331,0.0222254
Levant_Natufian,0.020488,0.1431895,-0.0377125,-0.1387295,0.030775,-0.079484,-0.025616,-0.0175375,0.114329,0.002005,0.0332085,-0.0222555,0.076486,0.002133,0.0153365,0.009016,-0.0154505,-0.001014,-0.02206,0.040832,0.001497,0.0001235,-0.003636,-0.0044585,0.006287
Nganassan,0.0476917,-0.4066181,0.1557885,0.0023902,-0.1594452,-0.0882129,0.0285066,0.0433367,0.0310876,0.0128477, 0.1028569,0.0094115,-0.0040734,-0.0261619,-0.0219731,-0.0123307,-0.0010952,0.0134165,0.0268365,-0.0008505,0.0431363,-0.0118954,0.0336096,0.0003977,0.0135556
West_African:Yoruba,-0.6300625,0.0625011,0.022113,0.0167079,0.0005035,0 .0124741,-0.044417,0.0477673,-0.0488813,0.0327694,0.0046205,0.0007904,0.0230561, 0.0009509,0.0125232,-0.0096067,0.0070763,0.0004491,0.006022,-0.00299,0.0015542,0.0023156,-0.0017592,-0.0004711,-0.0004246
Indian:IND_Great_Andamanese_100BP,-0.018212,-0.235603,-0.128598,0.098838,0.023389,-0.003904,-0.016686,-0.003231,0.059721,0.02041,0.023222,0.004646,-0.0055,0.001239,-0.016151,-0.008884,0.009518,0.001014,-0.004902,0.026888,-0.00549,0.011994,-0.013311,-0.001205,0.000958
Yamnaya_UKR,0.1160995,0.0878435,0.0465745,0.106106 ,-0.030467,0.038766,0.0074025,-0.0026535,-0.0502105,-0.070252,-0.002923,0.003297,-0.002453,-0.019198,0.0311475,0.013988,-0.0009775,-0.0118455,-0.0036455,0.001313,-0.001248,-0.001175,-0.003636,0.0211475,0.001856
IberomaurusianMAR_Taforalt,-0.189857,0.0814452,-0.0242866,-0.085595,0.027636,-0.0552202,-0.0705968,0.0184146,0.155397,0.003499,0.0209156,-0.0318316,0.0747168,-0.0513334,0.0711988,-0.0363032,0.0052676,-0.066106,-0.1424162,0.0389938,-0.0376836,-0.1255322,0.0730118,-0.0137606,0.0164534
RUS_Devils_Gate_Cave_N,0.0227646,-0.4480516,0.0728598,-0.052972,-0.0414846,-0.0407738,0.005264,0.011215,0.0075266,0.021431,-0.0507624,-0.005665,0.0027948,0.0100466,-0.0100162,-0.0120658,-0.0043026,0.0060558,0.0153102,0.0099798,0.0104566,-0.0287366,-0.0196458,-0.0016628,-0.0099152

[Jana]

Try these ancient components model for you and your mom. You don't need to post the results here.
If that CHG is not contained within Yamnaya, then maybe there is something interesting that happened.
I know for example that my father scores more Caucasus-related ancestry than the rest of us, but his is contained within the steppe component.
Hence, it is unlikely that it was mediated via a source from Caucasus. Maybe that's the same in your case.



Anatolia_Barcin_N,0.1175998,0.180118,0.0035312,-0.101158,0.0510443,-0.0483875,-0.0043582,-0.0069334,0.0362287,0.0807473,0.0079718,0.0118803,-0.0234545,0.0004691,-0.0419807,-0.0101913,0.0233091,0.0019866,0.0136954,-0.0097489,-0.0142249,0.0057723,-0.0041232,-0.0031658,-0.0043437
GEO_CHG,0.091058,0.102568,-0.083344,-0.00323,-0.08617,0.020638,0.024911,-0.001846,-0.128236,-0.074717,-0.006333,0.023979,-0.054856,0.004404,0.026601,-0.03275,0.02386,-0.013429,-0.022249,0.034767,0.033815,-0.007048,0.006532,-0.025787,-0.002036
WHG,0.1246365,0.116278,0.184789,0.189279,0.1546445 ,0.0464355,0.0131605,0.0372675,0.0890705,0.017768,-0.0153455,-0.015811,0.0159065,-0.0030275,0.053338,0.0582065,0.00502,0.016343,-0.0093015,0.055589,0.0944585,0.0111905,-0.049607,-0.160866,0.0170045
Baltic_LVA_HG,0.1292603,0.1004104,0.1802636,0.1972 329,0.1064236,0.0560919,0.0083574,0.0207106,0.0524 476,-0.0261394,-0.0046788,-0.0181151,0.0281434,-0.0053759,0.0355759,0.0480141,0.003072,0.0040856,-0.0059236,0.042325,0.0613526,0.0130531,-0.0262979,-0.1110549,0.0095949
IRN_Ganj_Dareh_N,0.0430252,0.0664158,-0.1550722,0.0047158,-0.122669,0.0235384,0.017109,-0.0011998,-0.082546,-0.0544158,-0.0028258,-0.0016186,0.0044896,-0.0062756,0.0316498,0.0561384,-0.0054242,0.0068664,0.0136508,-0.0334162,0.00856,-0.028836,-0.0110678,-0.039331,0.0222254
Levant_Natufian,0.020488,0.1431895,-0.0377125,-0.1387295,0.030775,-0.079484,-0.025616,-0.0175375,0.114329,0.002005,0.0332085,-0.0222555,0.076486,0.002133,0.0153365,0.009016,-0.0154505,-0.001014,-0.02206,0.040832,0.001497,0.0001235,-0.003636,-0.0044585,0.006287
Nganassan,0.0476917,-0.4066181,0.1557885,0.0023902,-0.1594452,-0.0882129,0.0285066,0.0433367,0.0310876,0.0128477, 0.1028569,0.0094115,-0.0040734,-0.0261619,-0.0219731,-0.0123307,-0.0010952,0.0134165,0.0268365,-0.0008505,0.0431363,-0.0118954,0.0336096,0.0003977,0.0135556
West_African:Yoruba,-0.6300625,0.0625011,0.022113,0.0167079,0.0005035,0 .0124741,-0.044417,0.0477673,-0.0488813,0.0327694,0.0046205,0.0007904,0.0230561, 0.0009509,0.0125232,-0.0096067,0.0070763,0.0004491,0.006022,-0.00299,0.0015542,0.0023156,-0.0017592,-0.0004711,-0.0004246
Indian:IND_Great_Andamanese_100BP,-0.018212,-0.235603,-0.128598,0.098838,0.023389,-0.003904,-0.016686,-0.003231,0.059721,0.02041,0.023222,0.004646,-0.0055,0.001239,-0.016151,-0.008884,0.009518,0.001014,-0.004902,0.026888,-0.00549,0.011994,-0.013311,-0.001205,0.000958
Yamnaya_UKR,0.1160995,0.0878435,0.0465745,0.106106 ,-0.030467,0.038766,0.0074025,-0.0026535,-0.0502105,-0.070252,-0.002923,0.003297,-0.002453,-0.019198,0.0311475,0.013988,-0.0009775,-0.0118455,-0.0036455,0.001313,-0.001248,-0.001175,-0.003636,0.0211475,0.001856
IberomaurusianMAR_Taforalt,-0.189857,0.0814452,-0.0242866,-0.085595,0.027636,-0.0552202,-0.0705968,0.0184146,0.155397,0.003499,0.0209156,-0.0318316,0.0747168,-0.0513334,0.0711988,-0.0363032,0.0052676,-0.066106,-0.1424162,0.0389938,-0.0376836,-0.1255322,0.0730118,-0.0137606,0.0164534
RUS_Devils_Gate_Cave_N,0.0227646,-0.4480516,0.0728598,-0.052972,-0.0414846,-0.0407738,0.005264,0.011215,0.0075266,0.021431,-0.0507624,-0.005665,0.0027948,0.0100466,-0.0100162,-0.0120658,-0.0043026,0.0060558,0.0153102,0.0099798,0.0104566,-0.0287366,-0.0196458,-0.0016628,-0.0099152


I only have mine G25 components. Fit is horrible...


Target: Feiichy
Distance: 4.0861% / 0.04086131

43.2 Anatolia_Barcin_N
38.4 Yamnaya_UKR
17.8 Baltic_LVA_HG
0.6 GEO_CHG

[Jana]

Any idea where his MENA (near eastern) ancestry is coming from? Byzantine from Vlach/Romanian admixture in Szeklers, or something else?
It's quite a lot for central European, I don't score any.

And his Uralic input is quite significant also for Magyar.