PDA

View Full Version : Eurogenes K9 Teal PCAs/Graphs/Dendograms



Kurd
04-12-2015, 04:22 AM
I started this thread for posting the various charts & graphs that are based on Eurogenes K9 'Teal". I will be uploading various charts & plots on a regular basis. The existence of a K9 spreadsheet should be helpful in the analysis. I will also be uploading a few graphs showing members' admix similarity to some K9 reference Asian groups.

This graph shows the admixture component distribution of the members who posted their results:


http://i.imgur.com/ADy2r17.jpg

E Asian= E Asian+ Amerindian+ Siberian
S Asian includes the little bit of Oceanian (based on FST) some members have. It was not worth creating a separate category for Oceanian.

kenji.aryan
04-12-2015, 05:38 AM
I think you swapped P gill and mcninja results so double check it, i can be wrong.

Kurd
04-12-2015, 05:54 AM
I think you swapped P gill and mcninja results so double check it, i can be wrong.

Good catch, but Gill's results are directly from McNinja's spreadsheet. McNinja's C Asian & S Asian somehow got switched, but on 2nd observation comparing McNinja's post with his spreadsheet for Gill, I noticed that Gill's & McNinja's results are the same. Does McNinja go by Gill as well?

Kurd
04-12-2015, 06:07 AM
The 1st graph shows the admixture similarity of the community to Brahuis, and the 2nd to Afghan Pashtuns. The Brahui and Pashtun references were formed using the average of the individuals on the K9 spreadsheet only. I will be going down the list and uploading community similarity to various Asian populations.

I will also start working on the PCAs tomorrow.

http://i.imgur.com/1k9WIsc.jpg
http://i.imgur.com/8XGtnzh.jpg

kenji.aryan
04-12-2015, 01:32 PM
Good catch, but Gill's results are directly from McNinja's spreadsheet. McNinja's C Asian & S Asian somehow got switched, but on 2nd observation comparing McNinja's post with his spreadsheet for Gill, I noticed that Gill's & McNinja's results are the same. Does McNinja go by Gill as well?

Yes on spreadsheet he is HRP0349.

Kurd
04-12-2015, 03:08 PM
Another reason why I decided to graph similarity to K9 spreadsheet populations is that the results from the K9 Oracle using the Gaussian method are quite different from the results using the least squares method. By comparing your oracle results to the admixture similarity graphs, you can surmise which results are more accurate.

http://i.imgur.com/aYKlymK.jpg
http://i.imgur.com/rLvy7ZH.jpg

Dr_McNinja
04-12-2015, 03:11 PM
Paulgill and I are both Punjabi Jatt Gills, I'm usually marked as Punjabi Jatt Muslim (HRP0349) Gill/Pansota and he's Punjabi Jatt Sikh (P.G.)

Dr_McNinja
04-12-2015, 03:13 PM
https://docs.google.com/spreadsheet/ccc?key=0AuXBmvmgdkfVdFMtRHVlZDBuQ3lMcjhxMDE4V3JoY lE#gid=54

I don't have Paulgill on this one. Did he post his K9 result somewhere?

Kurd
04-12-2015, 03:15 PM
Paulgill and I are both Punjabi Jatt Gills, I'm usually marked as Punjabi Jatt Muslim (HRP0349) Gill/Pansota and he's Punjabi Jatt Sikh (P.G.)

Thanks for clarifying. Moving forward I will make the adjustment on the charts. On the ones I have posted so far, your results are marked PGill by mistake.

Kurd
04-12-2015, 03:16 PM
https://docs.google.com/spreadsheet/ccc?key=0AuXBmvmgdkfVdFMtRHVlZDBuQ3lMcjhxMDE4V3JoY lE#gid=54

I don't have Paulgill on this one. Did he post his K9 result somewhere?

My mistake, I assumed that "Gill" on your spreadsheet was the same as PGill

Kurd
04-13-2015, 04:01 AM
Here is a plot of PC1 & PC2 with a few K9 spreadsheet populations. PC1/PC2 captures about 66% of the admixture variation. When combined with PC/PC4, which I will try to post in a couple of days, about 96% of the variation is captured. In other words, you can't infer your total distance from other members and K9 references by just looking at PC1/PC2 by itself. You would have to add your distance from PC3/PC4 to the distance in PC1/PC2.

The eigenvectors give you an idea of the various component axis. The component reductions are as they were in my previous post. I will plot another PCA later with more populations added.


http://i.imgur.com/gfcGwch.jpg
http://i.imgur.com/dWMgrKD.jpg
http://i.imgur.com/G8HWFGU.jpg

MonkeyDLuffy
04-13-2015, 04:10 AM
Just a quick question, is Kalyan me? :)

Kurd
04-13-2015, 04:12 AM
Just a quick question, is Kalyan me? :)

Kalyan is Punjabi Ramgarhia (Kalyan/Sembhi) (Anandpur Sahib) from Dr McNinja's spreadsheet

MonkeyDLuffy
04-13-2015, 04:13 AM
Yup that's me. Thanks!

Kurd
04-13-2015, 05:05 AM
Another reason why I decided to graph similarity to K9 spreadsheet populations is that the results from the K9 Oracle using the Gaussian method are quite different from the results using the least squares method. By comparing your oracle results to the admixture similarity graphs, you can surmise which results are more accurate.

http://i.imgur.com/aYKlymK.jpg
http://i.imgur.com/rLvy7ZH.jpg
http://i.imgur.com/HqqSFSd.jpg

Jessie
04-13-2015, 05:44 AM
Thanks for all this work Kurd. It is very interesting.

Kurd
04-15-2015, 04:25 AM
This PCA is updated with a few more K9 populations. Members' positions may have shifted a little from the previous PCA due to the addition of other populations. Refer to my previous post for explanations.

http://i.imgur.com/wJWLwbS.jpg
http://i.imgur.com/kEt6i4i.jpg
http://i.imgur.com/lnwypWG.jpg

Kurd
04-17-2015, 05:14 PM
At the request of a member, here is the community similarity graph to Tajiks. As per Rukha, these are Yunasb. Tajikstan Tajiks. For constructing the Tajik reference, I averaged the 13 Tajiks on the K9 spreadsheet.

http://i.imgur.com/hIpQiZ4.jpg

Kaido
04-17-2015, 05:34 PM
Where is PashtunD from?

Kurd
04-17-2015, 07:09 PM
Where is PashtunD from?

He is a Durrani from Kandhar. I am not sure if he is related to Zahra

Kurd
04-19-2015, 03:15 AM
At the request of a member, here is the community similarity graph to Tajiks. As per Rukha, these are Yunasb. Tajikstan Tajiks. For constructing the Tajik reference, I averaged the 13 Tajiks on the K9 spreadsheet.

http://i.imgur.com/hIpQiZ4.jpg

At the request of a member here is the similarity graph to Tajik Pamiri. The T Pamiri reference is comprised of the average of 8 T Pamiri individulas from the K9 spreadsheet.

http://i.imgur.com/UPO1Zor.jpg

MonkeyDLuffy
04-19-2015, 03:30 AM
Thank you kurd for making graphs and plots, they make it much easier to understand the differences.

Kurd
04-19-2015, 05:52 PM
The first graph shows the community similarity to Cambodians. The average of 8 K9 spreadsheet Cambodians was used as the reference. For the second the average of 15 K9 spreadsheet Bedouins was used as a reference.

As a heads up the next 2 graphs to be uploaded are for similarity to Turks & English.


http://i.imgur.com/VV3OoUV.jpg

http://i.imgur.com/VaaRHnG.jpg

paulgill
04-19-2015, 08:52 PM
Good catch, but Gill's results are directly from McNinja's spreadsheet. McNinja's C Asian & S Asian somehow got switched, but on 2nd observation comparing McNinja's post with his spreadsheet for Gill, I noticed that Gill's & McNinja's results are the same. Does McNinja go by Gill as well? Gedmatch M065657 and HRP0299, if ever in doubt. Dr. McNinja had assigned my 2.37% W. Asian on one of his spreadsheets to Archaic African, that I have none of. Your graph did look very weird to me, also let me tell you that one of the K9 admixture results are very weird to begin with, it is a junk calculator, you have not included me in the other graphs, may I ask you why?

Kurd
04-19-2015, 09:17 PM
Gedmatch M065657 and HRP0299, if ever in doubt. Dr. McNinja had assigned my 2.37% W. Asian on one of his spreadsheets to Archaic African, that I have none of. Your graph did look very weird to me, also let me tell you that one of the K9 admixture results are very weird to begin with, it is a junk calculator, you have not included me in the other graphs, may I ask you why?

I am not sure what you mean. I still don't see you on Dr McNinja's K9 David's spreadsheet. You realize that this thread is dedicated to David's K9 Teal results and not any gedmatch eurogenes calculator?

Did David send your results, and if so, did you post them somewhere?

Also, specifically what are you referring to by "weird results"?

paulgill
04-19-2015, 09:27 PM
I am not sure what you mean. I still don't see you on Dr McNinja's K9 David's spreadsheet. You realize that this thread is dedicated to David's K9 Teal results and not any gedmatch eurogenes calculator?

Did David send your results, and if so, did you post them somewhere?

Also, specifically what are you referring to by "weird results"?It was in response to an earlier post and now I realize that it simply was a case of mistaken identity. I have no interest in David's results, he will need to find someone else to sell his mumbo jumbo to.

Kurd
04-20-2015, 03:28 AM
At the request of community members, the first graph is for community similarity to Uzbeks, and the second to Turks. Uzbek reference was the average of around 20 Uzbeks, and the Turk reference was the average of 50+ Turks from various parts of Turkey. The results did not look accurate on the English graph, so I am re-doing it, and shall post later.

Also, I am working on a new type of graph, which will incorporate the similarity percentages for each member for various ethnic groups. I shall upload within the next couple of days.

http://i.imgur.com/mdQ4GUU.jpg

http://i.imgur.com/AMvO7x3.jpg

Kurd
04-20-2015, 04:23 AM
The following graphs summarizes/visualizes the members' cumulative similarities to various ethnic groups

http://i.imgur.com/sVSUZ4K.jpg

Chad Rohlfsen
04-20-2015, 04:32 AM
Kurd,

Here are my results. Could you throw me in as well, please? Thanks!

Oceanian 1E-005
European 0.518274
EEF 0.314541
East_Asian 1E-005
South_Asian 1E-005
Central Asian 0.151058
Amerindian 1E-005
Siberian 0.016077
Sub-Saharan 1E-005

Kurd
04-20-2015, 04:37 AM
Kurd,

Here are my results. Could you throw me in as well, please? Thanks!

Oceanian 1E-005
European 0.518274
EEF 0.314541
East_Asian 1E-005
South_Asian 1E-005
Central Asian 0.151058
Amerindian 1E-005
Siberian 0.016077
Sub-Saharan 1E-005

Ok, I will add you to future graphs, but will not be able to add you to graphs such as the one just posted because it requires re-doing all past graphs, which is way too time consuming.

Chad Rohlfsen
04-20-2015, 04:46 AM
Ok, I will add you to future graphs, but will not be able to add you to graphs such as the one just posted because it requires re-doing all past graphs, which is way too time consuming.

No problem! Thanks!

SwampThing27
04-20-2015, 04:47 AM
I'm really interested in seeing the English comparison graph if you are able to get it working.

Kurd
04-20-2015, 04:53 AM
I'm really interested in seeing the English comparison graph if you are able to get it working.

It should be uploaded within the next day or two

Kurd
04-20-2015, 05:02 AM
The following graphs summarizes/visualizes the members' cumulative similarities to various ethnic groups

http://i.imgur.com/sVSUZ4K.jpg

Cumulative similarity simply means that the similarity to various ehnic groups from previous graphs has been added. So for example Zahra is most similar to the combination of the ethnic groups listed

Helgenes50
04-20-2015, 05:13 AM
Cumulative similarity simply means that the similarity to various ehnic groups from previous graphs has been added. So for example Zahra is most similar to the combination of the ethnic groups listed

And in my case the less similar.
Thanks for your work.

Kurd
04-20-2015, 05:20 AM
And in my case the less similar.
Thanks for your work.

That may be the case for these groups, which are primarily SC Asia, Iranian, and SE Asian. If the groups were European, then the tables would be reversed, and we would be trading places with you.

Anabasis
04-20-2015, 05:55 AM
Kurd Thanks very much for that excellent work. Could you do those similarities with, Cypriots, Greeks, Armenians, Druzis, Serbians, Bulgarians, Estonians, (form mediterenians to balkans and baltic ?)

Kurd
04-20-2015, 05:15 PM
Kurd Thanks very much for that excellent work. Could you do those similarities with, Cypriots, Greeks, Armenians, Druzis, Serbians, Bulgarians, Estonians, (form mediterenians to balkans and baltic ?)

Pick one that is your favorite as it can take me up to 30 mins for me to do a graph from the analysis stage to graphing to color coding the ethnic groups to uploading and posting :)

Chad Rohlfsen
04-20-2015, 10:27 PM
Pick one that is your favorite as it can take me up to 30 mins for me to do a graph from the analysis stage to graphing to color coding the ethnic groups to uploading and posting :)

Personally, I'm interested in English, Basque, Sardinian, Norwegian, Lithuanian, and Greek. Those should be interesting.

Kurd
04-20-2015, 11:31 PM
Personally, I'm interested in English, Basque, Sardinian, Norwegian, Lithuanian, and Greek. Those should be interesting.

English is next followed by either Armenian or Greek

MonkeyDLuffy
04-21-2015, 03:14 AM
I am also interested in east europeans, can you do one on baltic too? Thanks :)

MonkeyDLuffy
04-21-2015, 04:43 AM
It was in response to an earlier post and now I realize that it simply was a case of mistaken identity. I have no interest in David's results, he will need to find someone else to sell his mumbo jumbo to.

Mumbo jambo? Can you please explain that paji? :)

paulgill
04-21-2015, 05:20 AM
Mumbo jambo? Can you please explain that paji? :)Google for the definition. The aDna changes drastically just in one generation as evident in your own
case, so it definitely not of great importance, at least not to me. There are tons of free tools available to analyze it, as at gedmatch etc., so none need to pay for another such tool just because it gives you different components or their ratios, because adjustments may be based on biased or mistaken assumptions.

MonkeyDLuffy
04-21-2015, 05:10 PM
Google for the definition. The aDna changes drastically just in one generation as evident in your own
case, so it definitely not of great importance, at least not to me. There are tons of free tools available to analyze it, as at gedmatch etc., so none need to pay for another such tool just because it gives you different components or their ratios, because adjustments may be based on biased or mistaken assumptions.

What does that mean? What changed in my case

paulgill
04-21-2015, 09:37 PM
What does that mean? What changed in my case
You are unlike a regular Punjabi Ramgarhia, you seem to be closer to Punjabi Jatts and the reason is that your Grandfather was a Baloch Tarkhan, as you mentioned somewhere. That Baloch aDna input have now moved you away from your own Baloch or Punjabi Tarkhan ethnic groups closer to Punjabi Jatts, etc. Yet your Grandfather's Ydna Baloch lineage is passed on to you unaffected, and you are a Baloch for that very reason, even today.

MonkeyDLuffy
04-21-2015, 09:42 PM
For some reasons i started to believe that the david's calculators are kind of biased.

Dr_McNinja
04-21-2015, 10:24 PM
Paulgill means autosomal DNA isn't as useful for deep ancestry (beyond familial relations), but it's useful for anthropological purposes when you do autosomal analyses of entire populations, as David does. Paying for individual analyses is a form of crowdfunding or donations so David can continue his other analyses.

When we do analyses of, for example, all North Indians, we can see a pattern autosomally that we can't through any other means (i.e, WHG peaking at high levels in Haryana).

MonkeyDLuffy
04-21-2015, 10:36 PM
You are unlike a regular Punjabi Ramgarhia, you seem to be closer to Punjabi Jatts and the reason is that your Grandfather was a Baloch Tarkhan, as you mentioned somewhere. That Baloch aDna input have now moved you away from your own Baloch or Punjabi Tarkhan ethnic groups closer to Punjabi Jatts, etc. Yet your Grandfather's Ydna Baloch lineage is passed on to you unaffected, and you are a Baloch for that very reason, even today.

Actually i have the least baloch component in comparison to other ramgarhias. On 23&me lots of ramgarhias show middle eastern mix while i have none. But I get what you are saying. Thats why i want to take another test, as i am not satisfied with 23&me.

Kurd
04-23-2015, 11:21 AM
The following graphs summarizes/visualizes the members' cumulative similarities to various ethnic groups

http://i.imgur.com/sVSUZ4K.jpg

The following 2 graphs show the community admixture similarity to Armenians and English. The Armenian and English references comprise the averages of the respective populations in the K9 Teal spreadsheet. I will be uploading a similarity graph to a S Asian group in the next couple of days.

http://i.imgur.com/KGZV8i5.jpg
http://i.imgur.com/sUkCAFw.jpg

EDIT: Just noticed that C Rohlfsen is not on the graphs. I will be revising and posting.

Chad Rohlfsen
04-23-2015, 10:25 PM
Kurd,

Is there any chance you could throw me in there? Thanks!

MonkeyDLuffy
04-23-2015, 11:11 PM
Who is the grewal guy? Sapporo?

Sapporo
04-23-2015, 11:29 PM
Who is the grewal guy? Sapporo?

Yes, it is me. :)

Kurd
04-24-2015, 12:26 PM
Kurd,

Is there any chance you could throw me in there? Thanks!

I have added you to the Armenian graph. Unfortunately, I was not able to add you to the English graph, since I had already deleted that sheet, as is the case most of the time. You will be on future graphs, however. That is one of the reasons members need to post their data timely if they want to be included in a project.

http://i.imgur.com/bFcaEfC.jpg

John Doe
04-24-2015, 01:46 PM
Here are my results:

K9

Pop ID Oceanian European EEF East_Asian South_Asian Central_Asian Amerindian Siberian Sub-Saharan
Guy_Jacks Guy_Jacks 1E-005 0.192158 0.520595 0.007275 1E-005 0.256318 0.000144 0.007463 0.016027

K9 Oracle
1- [email protected] 1,79493986096744
2- [email protected] 1,96549842604872
3- [email protected] 2,68213368716649
4- [email protected] 3,3995601884196
5- [email protected] 3,41132532926224
6- [email protected] 5,60260583573617
7- [email protected] 6,98255882008433
8- [email protected] 7,61575233435749
9- [email protected] 8,40370446762553
10- [email protected] 10,4793327858801
11- [email protected] 10,7799141811867
12- [email protected] 11,877869150065
13- [email protected] 13,0573894220114
14- [email protected] 14,4041602206348
15- [email protected] 14,7199733346248
16- [email protected] 15,427184679882
17- [email protected] 15,484295955082
18- [email protected] 17,3793707113711
19- [email protected] 19,019100010487
20- [email protected] 19,2165968795965
21- [email protected] 20,40412992713
22- [email protected] 20,5885567716558
23- [email protected] 20,8309778082575

Can you add me too? Thanks.

The Barnacle
04-24-2015, 02:32 PM
John your there already, no?

John Doe
04-24-2015, 02:42 PM
John your there already, no?

Well, there's someone called Doe on the graph, but I'm not entirely sure it's me.

Edit: Yeah yeah, it has to be me.

The Barnacle
04-24-2015, 03:59 PM
Kurd quick question, are the afghan Pashtuns in this experiment the 4 metspalu Northern Pashtuns?

Kurd
04-24-2015, 06:23 PM
Kurd quick question, are the afghan Pashtuns in this experiment the 4 metspalu Northern Pashtuns?

I'm not really sure

Kurd
04-24-2015, 08:28 PM
I started this thread for posting the various charts & graphs that are based on Eurogenes K9 'Teal". I will be uploading various charts & plots on a regular basis. The existence of a K9 spreadsheet should be helpful in the analysis. I will also be uploading a few graphs showing members' admix similarity to some K9 reference Asian groups.

This graph shows the admixture component distribution of the members who posted their results:


http://i.imgur.com/ADy2r17.jpg

E Asian= E Asian+ Amerindian+ Siberian
S Asian includes the little bit of Oceanian (based on FST) some members have. It was not worth creating a separate category for Oceanian.

The following graph compares members' admixture to 45ky Ust Ishim from Siberia, the oldest modern human genome to be sequenced

http://i.imgur.com/IkO2CMy.jpg

MonkeyDLuffy
04-24-2015, 11:05 PM
What was the mixture of ust ishim?

kenji.aryan
04-25-2015, 01:44 AM
..............

Generalissimo
04-25-2015, 02:28 AM
I wonder why i share more similarity with Ust ishim considering my largest & total segment at 400 SNPs with him is only 3.5 cM .


You don't share any IBD segments of 3.5 cM with Ust-Ishim. No one does. Ust-Ishim is too old.

You might share some IBD segments with Ust-Ishim that are much less than 1 cM, but to find them you'd need a full genome sequence analysis that relies on very rare alleles to find extremely old IBD.

parasar
04-25-2015, 02:39 AM
What was the mixture of ust ishim?

He is an early modern human from whose type most Eurasians are descended. He is just after the basal Eurasian split on the side that mixed (slightly) with Neanderthals.

Dr_McNinja
04-25-2015, 03:31 AM
The following graph compares members' admixture to 45ky Ust Ishim from Siberia, the oldest modern human genome to be sequenced

http://i.imgur.com/IkO2CMy.jpgI wonder why that Pashtun is up there with us. What's Ust-Ushim's K9 breakdown?

Kurd
04-25-2015, 03:54 AM
You don't share any IBD segments of 3.5 cM with Ust-Ishim. No one does. Ust-Ishim is too old.

You might share some IBD segments with Ust-Ishim that are much less than 1 cM, but to find them you'd need a full genome sequence analysis that relies on very rare alleles to find extremely old IBD.

I believe he meant IBS segments.

Mine happen to be:

Minimum threshold size to be included in total = 200 SNPs
Mismatch-bunching Limit = 100 SNPs
Minimum segment cM to be included in total = 3.0 cM




Chr
Start Location
End Location
Centimorgans (cM)
SNPs




2




206373925




208989176




3.5




443






2




216508617




217875275




3.4




302






2




235558975




238234971




4.6




563






4




39925692




41491873




3.2




337






5




172705357




173734018




3.5




265






6




1543156




2701682




3.5




300






7




12118001




13901832




3.0




479






10




3133667




3776476




3.3




204






10




33825481




39098677




3.4




676






11




130837777




132320578




4.2




453






20




318150




929418




3.2




240






20




57526824




58492611




4.0




243




Largest segment = 4.6 cM
Total of segments > 3 cM = 42.9 cM

Kurd
04-25-2015, 04:04 AM
I wonder why that Pashtun is up there with us. What's Ust-Ushim's K9 breakdown?



Oceanian
EEF
European
Amerindian
Central_Asian
South_Asian
East_Asian
Siberian
Sub-Saharan


12.7%
12.5%
8.4%
2.1%
1.5%
32.9%
13.6%
4.5%
11.8%

Kurd
04-25-2015, 04:28 AM
What was the mixture of ust ishim?

Ust Ishim belonged to the first wave of humans to migrate out of Africa into Europe and Asia, before or at the time of the East West Eurasian split. Admixture wise he is closest to S Asian populations, although he has higher SSA than S Asians. This is consistent with him living in a time period when humans had recently migrated out of Africa, relatively speaking. If I had to guess, I would say physically he probably resembled certain SE Asian populations, based on his environment and time since OOA.

Based on his Y DNA (I believe a subclade of LT (K2)) and mtDNA of R) he probably originated from the India region.

Genetically he is closer to modern East Asians than to Europeans. Modern Europeans are believed to have derived their ancestry from three ancient groups, WHG who lived 45,000 years ago and most likely originated in the second human migration out of Africa into Europe, EEF from about 9,000 years ago, and ANE from the steppe herders of northwestern Asia.

Some believe that Ust Ishim's group may have become extinct during the ice age.

kenji.aryan
04-25-2015, 06:21 AM
...........

Chad Rohlfsen
04-25-2015, 03:51 PM
Anything that is a mix of West and East Eurasian (to make Crown Eurasian), will be high in Ust-Ishim type of ancestry. I tested him as a component in a supervised run, and some Siberians scored 85% Ust_Ishim. I wouldn't make anything of it, at this point.

Dr_McNinja
04-25-2015, 07:33 PM
Oceanian
EEF
European
Amerindian
Central_Asian
South_Asian
East_Asian
Siberian
Sub-Saharan


12.7%
12.5%
8.4%
2.1%
1.5%
32.9%
13.6%
4.5%
11.8%



Ah, the EEF explains it.

Kurd
04-26-2015, 02:23 AM
The following is the cluster analysis using the Euclidean method for distance measurement (pairing you with your neighbor), and the "furthest neighbor" method for clusteroid linkage. If you need more info on cluster analysis, see my posts on the subject under the MDLP K23 thread I started about a month ago.

http://i.imgur.com/Yw7Z2yU.jpg

Here is the clusteroid info

http://i.imgur.com/QclaHm6.jpg

MonkeyDLuffy
04-26-2015, 06:29 AM
What is most representative observation?

thrax
04-26-2015, 07:16 AM
I'd like to be in these dendrograms as well.

Kurd
04-26-2015, 12:17 PM
What is most representative observation?

It is the member that is closest to the average/median of the cluster. Least representative is the member that is furthest from the median of the cluster