Page 29 of 29 FirstFirst ... 19272829
Results 281 to 287 of 287

Thread: Regional G25 Neolithic vs EBA Britain

  1. #281
    Registered Users
    Posts
    301
    Sex
    Nationality
    Scottish
    mtDNA
    J1c8 / L1b

    Scotland England
    I ran the K11 with default pen at 10,000 cycles.

    K11 Northern Europe

    Irish_&_Scottish_&_Welsh,84.2
    French_North_Central,5.6
    French_Northeast_&_Belgian_&_German_West,3
    German_Northwest_&_Swedish_South,2.2
    Norwegian_&_Swedish_North_Central,2.2
    Finnish,1.2
    Estonian_Latvian_Lithuanian,0.4
    French_East_&_German_South_&_Austrian,0.4
    French_South_&_French_Basque,0.4
    German_East_&_Czech_&_Austrian,0.2
    Polish_Belarusian_Ukrainian,0.2


    EDIT: Just finished running the K16 with default pen and 10,000 cycles.

    K16 Europe

    Irish_Scottish_Welsh,85.8
    French_Belgian,7.2
    Norwegian_Swedish,2.4
    German_Czech_Austrian,1.8
    Finnish,1.6
    Albanian_Greek,0.4
    Saami,0.4
    Basque,0.2
    Portuguese_Spanish,0.2
    Last edited by Ais; 12-08-2018 at 10:22 AM.
    DNA Tribes
    Me
    92.1% European, 3.2% South Asian, 2.9% North African & Middle Eastern, 0.9% Siberian, 0.6% Amerindian, 0.2% Sub-Saharan African

    My mother
    95% European, 2.9% North African & Middle Eastern, 2.1% Amerindian

  2. The Following User Says Thank You to Ais For This Useful Post:

     Finn (12-09-2018)

  3. #282
    Gold Member Class
    Posts
    1,176
    Sex
    Location
    Florida, USA.
    Ethnicity
    English, Scottish & Irish
    Nationality
    American
    Y-DNA
    I-A13243
    mtDNA
    H1e2

    England Scotland Ireland United States of America Vatican Germany Schleswig-Holstein
    Quote Originally Posted by Ais View Post
    I ran the K11 with default pen at 10,000 cycles.

    K11 Northern Europe

    Irish_&_Scottish_&_Welsh,84.2
    French_North_Central,5.6
    French_Northeast_&_Belgian_&_German_West,3
    German_Northwest_&_Swedish_South,2.2
    Norwegian_&_Swedish_North_Central,2.2
    Finnish,1.2
    Estonian_Latvian_Lithuanian,0.4
    French_East_&_German_South_&_Austrian,0.4
    French_South_&_French_Basque,0.4
    German_East_&_Czech_&_Austrian,0.2
    Polish_Belarusian_Ukrainian,0.2


    EDIT: Just finished running the K16 with default pen and 10,000 cycles.

    K16 Europe

    Irish_Scottish_Welsh,85.8
    French_Belgian,7.2
    Norwegian_Swedish,2.4
    German_Czech_Austrian,1.8
    Finnish,1.6
    Albanian_Greek,0.4
    Saami,0.4
    Basque,0.2
    Portuguese_Spanish,0.2
    Hello Ais,

    Just out of curiosity, how much difference are you seeing from a 10,000 cycle run and a shorter one? Say for example, 1000 cycle run or Poi’s default 200 cycles.
    Known Paper Trail: 45.3% English, 29.7% Scottish, 12.5% Irish, 6.25% German & 6.25% Italian. Or: 87.5% British Isles, 6.25% German & 6.25% Italian.
    LivingDNA: 88.1% British Isles (59.7% English, 27% Scottish & 1.3% Irish), 5.9% Europe South (Aegian 3.4%, Tuscany 1.3%, Sardinia 1.1%), 4.4% Europe NW (Scandinavia) & 1.6% Europe East, (Mordovia).
    FT Big Y: I1-Z140 branch I-F2642 >Y1966 >Y3649 >A13241 >Y3647 >A13248 (circa 930 AD) >A13242/YSEQ (circa 1075 AD) >A13243/YSEQ (circa 1660 AD).

  4. The Following User Says Thank You to JMcB For This Useful Post:

     Ais (12-09-2018)

  5. #283
    Registered Users
    Posts
    2,447
    Sex
    Location
    Groningen
    Ethnicity
    Northwest European
    Nationality
    NL
    Y-DNA
    E-V22

    Netherlands
    Quote Originally Posted by Capitalis View Post
    All of your posts revolve around your preconceptions based on the same coloured charts of the Netherlands. You never change your views no matter what the results of the samples are. I honestly find your posts that reply to any of my research to be confusing and worry that you will confuse any readers of my posts. This is very frustrating for me.

    All I can judge is that people here always "like" your posts, so I think this is the wrong forum for me because I put so much time into these posts with actual G25/nMonte3 results that I try to match with archaeology/history when possible and then they are always questioned by you.

    I base my posts on the results of G25/nMonte3 and PCA plots. You ask why the Anglo-Saxon samples score so much Nordic and low NW Germanic, only because they don't match your preconceptions - this is not my problem, this is the results of the calculator, which is determined by the ancestry of the actual sample. The calculators I make are good - David and Ger have done a marvellous job with G25/nMonte3; the calculators don't produce nonsensical results.

    In
    case you think my calculators are wrong, please look at David's PCA plot below and remember that the Anglo-Saxon samples are Nordic - they plot by modern Norwegians, not by modern NW Germans! Have you read (and absorbed) any of my posts? You might say yes but I say no.

    I don't like coming to Anthrogenica to argue why my research doesn't match your preconceptions, so it would be best not to ask me these questions in all honesty. But it would be better for me to leave because I've made the calculators I wanted and if people would run 10,000 cycles in R for themselves the results would be much better (more accurate, more informative) than the 200 cycles of the web runner.

    My life doesn't exist to argue; it is for research and interpreting results. This forum is sometimes more for supporting preconceptions - I've done that to but I try to admit my faults.

    Maybe my research has no value anyway but I'll leave it to others now.

    Very sorry.
    No they don't.

    My checkfit Global 25



    So no pre-concepts just outcomes of the G25 stuf.

  6. The Following User Says Thank You to Finn For This Useful Post:

     palamede (12-10-2018)

  7. #284
    Registered Users
    Posts
    301
    Sex
    Nationality
    Scottish
    mtDNA
    J1c8 / L1b

    Scotland England
    Quote Originally Posted by JMcB View Post
    Hello Ais,

    Just out of curiosity, how much difference are you seeing from a 10,000 cycle run and a shorter one? Say for example, 1000 cycle run or Poiís default 200 cycles.
    I find there can be a difference with default pen, but not as much with pen=0. There's a marked difference with the webrunner, but since that's only 200 cycles, I suppose it makes sense.

    Here's an example, my results in the webrunner for Capitalis' K16:

    Irish, Scottish & Welsh, 76.67
    German, Czech & Austrian, 8.33
    Norwegian & Swedish, 7.5
    French & Belgian, 6.67
    Finnish, 0.83

    It's quite different from the results in my previous post. Perhaps it's more accurate, since I don't receive trace amounts of Saami or Albanian Greek (although Greek does sometimes appear in my nMonte runs as well), but it's a big difference nonetheless.
    DNA Tribes
    Me
    92.1% European, 3.2% South Asian, 2.9% North African & Middle Eastern, 0.9% Siberian, 0.6% Amerindian, 0.2% Sub-Saharan African

    My mother
    95% European, 2.9% North African & Middle Eastern, 2.1% Amerindian

  8. The Following User Says Thank You to Ais For This Useful Post:

     palamede (12-10-2018)

  9. #285
    Gold Member Class
    Posts
    1,176
    Sex
    Location
    Florida, USA.
    Ethnicity
    English, Scottish & Irish
    Nationality
    American
    Y-DNA
    I-A13243
    mtDNA
    H1e2

    England Scotland Ireland United States of America Vatican Germany Schleswig-Holstein
    Quote Originally Posted by Ais View Post
    I find there can be a difference with default pen, but not as much with pen=0. There's a marked difference with the webrunner, but since that's only 200 cycles, I suppose it makes sense.

    Here's an example, my results in the webrunner for Capitalis' K16:

    Irish, Scottish & Welsh, 76.67
    German, Czech & Austrian, 8.33
    Norwegian & Swedish, 7.5
    French & Belgian, 6.67
    Finnish, 0.83

    It's quite different from the results in my previous post. Perhaps it's more accurate, since I don't receive trace amounts of Saami or Albanian Greek (although Greek does sometimes appear in my nMonte runs as well), but it's a big difference nonetheless.

    In my case, changing the cycles does switch things around a little but nothing radical. I tend to stick with Pen 0.001 because Huijbregt’s suggests that setting and if I remember correctly, Capitalis designed his calculators with that in mind.

    Be that as it may, here’s a comparison between 1000 & 200 cycles.


    Custom:AGUser_JMcB
    Pen 0.001

    1000 cycles
    500 batches

    Fit 0.8247

    German Northwest And Swedish South K8 36.8
    French Northeast And Belgian And German West K8 28.8
    Irish And Scottish And Welsh K8 20.8
    French South And French Basque K8 4.8
    Norwegian And Swedish North Central K8 4.4
    French North Central K8 2.4
    German East And Czech And Austrian K8 1.2
    French East And German South And Austrian K8 0.8




    Custom:AGUser_JMcB
    Pen 0.001

    200 cycles
    100 batches

    Fit 0.7027

    French Northeast And Belgian And German West K8 32.5
    German Northwest And Swedish South K8 26.67
    Irish And Scottish And Welsh K8 20.83
    Norwegian And Swedish North Central K8 8.33
    French South And French Basque K8 6.67
    French East And German South And Austrian K8 3.33
    German East And Czech And Austrian K8 1.67
    French North Central K8 0
    Known Paper Trail: 45.3% English, 29.7% Scottish, 12.5% Irish, 6.25% German & 6.25% Italian. Or: 87.5% British Isles, 6.25% German & 6.25% Italian.
    LivingDNA: 88.1% British Isles (59.7% English, 27% Scottish & 1.3% Irish), 5.9% Europe South (Aegian 3.4%, Tuscany 1.3%, Sardinia 1.1%), 4.4% Europe NW (Scandinavia) & 1.6% Europe East, (Mordovia).
    FT Big Y: I1-Z140 branch I-F2642 >Y1966 >Y3649 >A13241 >Y3647 >A13248 (circa 930 AD) >A13242/YSEQ (circa 1075 AD) >A13243/YSEQ (circa 1660 AD).

  10. The Following 2 Users Say Thank You to JMcB For This Useful Post:

     Finn (12-09-2018),  palamede (12-10-2018)

  11. #286
    Registered Users
    Posts
    231
    Sex
    Y-DNA
    J2a-M319

    Can someone tell me what penalization does? And how I can enable it using R?

  12. #287
    Registered Users
    Posts
    950
    Sex
    Location
    Lisbon, Portugal
    Ethnicity
    Northwest Iberian
    Nationality
    Portuguese
    Y-DNA
    E-Y31991 > PF4428
    mtDNA
    H20

    Portugal 1143 Portugal 1485 Portugal Order of Christ
    Quote Originally Posted by Sorcelow View Post
    Can someone tell me what penalization does? And how I can enable it using R?
    It's explained in Introduction nMonte3
    2. The nMonte results are overfitted models. In the first place the populations are related and therefore share common DNA. Secondly, nearly all populations are genetically equivalent to a mixture of other populations. For instance:
    German <- fp*French + (1-fp)*Polish. Because nMonte can itself choose the mixing parameter fp, this mixture will nearly always be preferred above true German.
    As a result of these two ways of overfitting, the estimates will be contaminated by noise; sometimes very much so.
    In machine learning, procedures directed at improving this situation are called 'regularization'.
    "Regularization artificially discourages complex or extreme explanations of the world even if they better fit what has been observed. The idea is that such explanations are unlikely to generalize well to the future; they may happen to explain a few data points from the past well, but this may just be because of accidents of the sample."(https://www.quora.com/What-is-regula...chine-learning)
    In nMonte3 the results are regularized by some degree of penalizing reference samples with great distances to the target. In other words, admixtures with small distances to the target are preferred to admixtures with large distances to the target.
    By default the degree of penalization is set at 0.001. The regularization can be switched off by adding an additional parameter pen=0 to the getMonte function, for instance: getMonte('myData.csv','myTarget.csv',pen=0)
    YDNA - E-Y31991>PF4428. Domingos Rodrigues, b. circa 1680 Hidden Content , Viana do Castelo, Portugal
    mtDNA - H20. Maria Josefa de Almeida, b. circa 1750 Hidden Content , Porto, Portugal

    Global25 PCA plot Hidden Content ; Western Europe-only Hidden Content
    Hidden Content

    [1] "distance%=2.2305"

    Ruderico

    Hallstatt_Bylany,43.4
    Iberia_BA,34.4
    Roman_Collegno_o1,18.4
    Iberomaurusian,3.8

  13. The Following 4 Users Say Thank You to Ruderico For This Useful Post:

     JMcB (02-12-2019),  Onur Dincer (02-13-2019),  Saetro (02-13-2019),  Sorcelow (02-12-2019)

Page 29 of 29 FirstFirst ... 19272829

Similar Threads

  1. K1a10a from the Neolithic in Britain
    By J1 DYS388=13 in forum K
    Replies: 2
    Last Post: 05-17-2018, 07:38 PM
  2. Population Replacement in Early Neolithic Britain
    By rozenfeld in forum Ancient (aDNA)
    Replies: 79
    Last Post: 02-28-2018, 09:14 PM
  3. % realy high in regional...
    By picto22 in forum Living DNA
    Replies: 77
    Last Post: 08-06-2017, 06:22 PM
  4. Regional Personality Differences in Great Britain
    By Jean M in forum Autosomal (auDNA)
    Replies: 1
    Last Post: 04-10-2015, 05:19 PM
  5. Replies: 1
    Last Post: 09-11-2014, 07:19 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •