Announcement

Collapse
No announcement yet.

Dodecad Ancestry Project

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    I'm FFD046

    Comment


    • #17
      29

      That looks amazing but where is 29?

      Comment


      • #18
        DOH it only labeled every other one. This one should be better. Sorry!
        Attached Files

        Comment


        • #19
          Thanks Kasa

          That is excellent

          Is Dienekes likely to split North European like Polako has?

          Originally posted by Kasandra View Post
          DOH it only labeled every other one. This one should be better. Sorry!

          Comment


          • #20
            On the ISOGG newsletter, it says that FTDNA's Population Finder looks at autosomes up to gt grandparents. Maybe this is why they can only give me 100% European. Where as Dienekes and Polako give a more detailed 'split'. Their analyses must be on pre gt grandparents.

            Just some thoughts

            Comment


            • #21
              Originally posted by Kasandra View Post
              DOH it only labeled every other one. This one should be better. Sorry!
              This is great! How did I get more Northern European than you brian? lol

              Comment


              • #22
                Originally posted by Salabencher View Post
                This is great! How did I get more Northern European than you brian? lol
                Inbreeding! Only joking

                Comment


                • #23
                  Thanks for the interesting picture Kasandra!

                  Comment


                  • #24
                    Just wanted to push this up again to show some of the results.

                    Currently Dienekes is experimenting with Mclust (a clustering software).
                    Its a way to find clusters (a cluster in there is a minimum of 3 people who are similiar) and assines people to the nearest cluster that exists at the moment.

                    The special about it is:
                    1. It puts you into one (closest) cluster, no interpreting about position on maps in different dimensions about where it belongs.
                    2. Mclust is able to have non-circular clusters.

                    Here is a map I did from the latest results.

                    It assined me to cluster 7.

                    Comment


                    • #25
                      Thanks Daniel

                      Is this only open to 23andMe folks?


                      Originally posted by Daniel72 View Post

                      It assined me to cluster 7.

                      [img]http://img84.imageshack.us/img84

                      Comment


                      • #26
                        Originally posted by briancowings View Post
                        Thanks Daniel

                        Is this only open to 23andMe folks?
                        No. FTDNA also.
                        But he currently closed the "Open ended" oppornuty to send in data. But he might open it again.

                        I will see to post here if this happens again.

                        Comment


                        • #27
                          Daniel72: Are the clusters shown on your map for K=64 results? If so, I am in Cluster 11 plotting as usual with Romanians despite my alleged Irish/Italian heritage.

                          Comment


                          • #28
                            Originally posted by varlene View Post
                            Daniel72: Are the clusters shown on your map for K=64 results? If so, I am in Cluster 11 plotting as usual with Romanians despite my alleged Irish/Italian heritage.
                            Yes, its from the K=64

                            It seems that all these kind of methods (MDS aswell as this clustering) has serious Problems with people who are of mixed decent.

                            they tend to end up in the middle of their parents clusters, wich is correct. But sometimes this space is already occupied by another cluster.

                            There is someone in the project, who is half Sicilian and half Polish. He ends up in Cluster 11 (Balkan) aswell.

                            And Cluster 11 turns out to be "Balkan" not "Romanian", because the Bulgars, Serbs and Croats all come out as Cluster 11 aswell.

                            Comment


                            • #29
                              I would love to know if I'm a 12 or a 7... or more remotely, a 33.

                              Comment


                              • #30
                                He has my FTDNA data, so will I be 'clustered'?



                                Originally posted by Daniel72 View Post
                                No. FTDNA also.
                                But he currently closed the "Open ended" oppornuty to send in data. But he might open it again.

                                I will see to post here if this happens again.

                                Comment

                                Working...
                                X