Skip to main content

TR Memescape

  • TalkRational: One of the five most chickenshit supposedly rationalist sites on the Internet.

Topic: Bootstrapping (Read 1138 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
  • Fenrir
Re: Bootstrapping
Reply #100
Soft polytomies are usually an artefact of insufficient  data. If Socrates is so energised about the science not progressing fast enough then he can always contribute new data. Just be careful which side is up when you measure it Socrates, getting the basics wrong could be embarrassing :)

Course that won't solve Socrates' dilemma, his issues occur in well supported nodes earlier on the tree.

Like anyone would let Socrates anywhere near an actual fossil.
It's what plants crave.

  • Doobie Keebler
  • Ridiculous Callipygous
Re: Bootstrapping
Reply #101
Hey Socrates, I'm curious.

Have you asked these experts precisely what they think this polytomy indicates? If so, could you reproduce it here?

Go ahead. Ask the study authors. Tell us what you find out.

If anyone wishes to take up this challenge, please do. The study authors have been loath to discuss it to this point. But I am certainly not wasting my time arguing this with you folks.
You yourselves can see the material in the Supplementary Information. You yourselves can contact the authors. If this is important to you that is what you might do. If this is not important to you, why are you wasting your time here? It is neither here not there to me what you do. If you can get an answer from the authors then we can all gain.
The authors do not acknowledge the significance of the huge polytomy. And you folks have not acknowledged the significance of the huge polytomy. I have explained the significance.

The authors do not even try to explain the significance of the huge polytomy. The significance of Euparaves (as I have called it for reference).

It's a yuuge polytomy, the very best polytomy, many people are saying so, why isn't the fake news talking about that? sad!
"I'm over 70 and have never seen such , arrogance, incompetence and Ill -intentions as this President and his aids."    The Dotard     (posted 12 days after his 68th birthday)

  • socrates1
Re: Bootstrapping
Reply #102
Hey Socrates, I'm curious.

Have you asked these experts precisely what they think this polytomy indicates? If so, could you reproduce it here?

Go ahead. Ask the study authors. Tell us what you find out.

If anyone wishes to take up this challenge, please do. The study authors have been loath to discuss it to this point. But I am certainly not wasting my time arguing this with you folks.
You yourselves can see the material in the Supplementary Information. You yourselves can contact the authors. If this is important to you that is what you might do. If this is not important to you, why are you wasting your time here? It is neither here not there to me what you do. If you can get an answer from the authors then we can all gain.
The authors do not acknowledge the significance of the huge polytomy. And you folks have not acknowledged the significance of the huge polytomy. I have explained the significance.

The authors do not even try to explain the significance of the huge polytomy. The significance of Euparaves (as I have called it for reference).

Does anyone understand the significance of that huge polytomy (Euparaves)? If so please explain it.
I should know better by now to ask such a question. Nobody will answer it.
It is better to remind you that such a huge polytomy means that the cladistic analysis cannot tell us anything about the relationship of the branches.
This is definitely not something that the authors have ever mentioned. Because if they did the imaginary nodes within Euparaves would disappear.
And they would be stuck with no evidence of a link between tyrannosauroids and Euparaves.
They would then be stuck having to explain that away.
  • Last Edit: June 18, 2017, 08:53:44 PM by socrates1

Re: Bootstrapping
Reply #103
Now I see the problem. You folks do not know this subject. I have been studying it over the course of the last year and having some discussion with experts. It is not reasonable for me to think you know it.

Not only that, I have the TNT software and have imported the matrix data from the two studies and run them. And then run the bootstrap and jackknife support calculations and got basically the same results as they did. Including the huge polytomy. It is not that I know better than them. It is that I am acknowledging the results that they themselves obtained that they do not acknowledge. But you yourselves can see all this from their own studies - look at the Supplementary Information.


Given your past record of being utterly incapable of understanding or performing the simplest of analyses, I strongly suspect this post of yours is complete bullshit.  But hey, prove me wrong.  Let's see some (any) evidence you did what you claim.

If you cannot provide such evidence, that's OK.

Re: Bootstrapping
Reply #104
Now I see the problem. You folks do not know this subject. I have been studying it over the course of the last year and having some discussion with experts. It is not reasonable for me to think you know it.
I, and many people here know you as someone who has demonstrated that he is unable to do really simple things, like for example interpret simple diagrams (or any picture really). To trust that you studying something for a year would result in understanding seems unwise.

  • socrates1
Re: Bootstrapping
Reply #105
Quote
The authors do not even try to explain the significance of the huge polytomy. The significance of Euparaves (as I have called it for reference).

Does anyone understand the significance of that huge polytomy (Euparaves)? If so please explain it.
I should know better by now to ask such a question. Nobody will answer it.
It is better to remind you that such a huge polytomy means that the cladistic analysis cannot tell us anything about the relationship of the branches.
This is definitely not something that the authors have ever mentioned. Because if they did the imaginary nodes within Euparaves would disappear.
And they would be stuck with no evidence of a link between tyrannosauroids and Euparaves.
They would then be stuck having to explain that away.

People here are beginning to acknowledge the existence of the huge polytomy. Perhaps someone will acknowledge the significance of it. But that may take a while.

  • VoxRat
  • wtactualf
Re: Bootstrapping
Reply #106
"I understand Donald Trump better than many people because I really am a lot like him." - Dave Hawkins

  • socrates1
Re: Bootstrapping
Reply #107
http://ib.berkeley.edu/courses/ib200a/labs/ib200a_lab10_bootstrap_jackknife_bremer.pdf
Bootstrapping calculates a support value for each node based on the fraction of samples that support that node. The highest support value is 100, while values below 70 are usually considered weak. Values below 50 aren't shown; in fact, branches below 50 are collapsed and shown as a polytomy.

Worth repeating.
The branches are collapsed because they are not supported. The analysis cannot tell the relationship between the branches. For example the cladistic analysis cannot determine if creatures like oviraptorids are ancestral or descendant to basal Paraves.
Authors continually fail to understand this basic point and put out unsupported cladograms.
If researchers and authors acknowledged this point they would then have to assess the evidence about where to place oviraptorids. The cladistic analysis does not help in that regard.

For example, the fact that the oviraptorids are dated tens of millions of years later (closer to today) than the basal Paraves would actually settle the issue.
  • Last Edit: June 19, 2017, 07:16:55 AM by socrates1

  • Faid
Re: Bootstrapping
Reply #108
94%.

Worth repeating.
Who even made the rule that we cannot group ducks and fish together for the simple reason that they are both aquatic? If I want to group them that way and it serves my purpose then I can jolly well do it however I want to and it is still a nested hierarchy and you can't tell me that it's not.

  • socrates1
Re: Bootstrapping
Reply #109
http://ib.berkeley.edu/courses/ib200a/labs/ib200a_lab10_bootstrap_jackknife_bremer.pdf
Bootstrapping calculates a support value for each node based on the fraction of samples that support that node. The highest support value is 100, while values below 70 are usually considered weak. Values below 50 aren't shown; in fact, branches below 50 are collapsed and shown as a polytomy.

Worth repeating.
The branches are collapsed because they are not supported. The analysis cannot tell the relationship between the branches. For example the cladistic analysis cannot determine if creatures like oviraptorids are ancestral or descendant to basal Paraves.
Authors continually fail to understand this basic point and put out unsupported cladograms.
If researchers and authors acknowledged this point they would then have to assess the evidence about where to place oviraptorids. The cladistic analysis does not help in that regard.

For example, the fact that the oviraptorids are dated tens of millions of years later (closer to today) than the basal Paraves would actually settle the issue.

I will be leaving this topic soon. I know you folks need advance warning.

  • VoxRat
  • wtactualf
Re: Bootstrapping
Reply #110
<  clip pointless self-quote >
<  clip pointless self-quote >

I will be leaving this topic soon.
I guess you've said what you felt you needed to say.
For whatever reason.

Bye now.
"I understand Donald Trump better than many people because I really am a lot like him." - Dave Hawkins

  • socrates1
Re: Bootstrapping
Reply #111
http://ib.berkeley.edu/courses/ib200a/labs/ib200a_lab10_bootstrap_jackknife_bremer.pdf
Bootstrapping calculates a support value for each node based on the fraction of samples that support that node. The highest support value is 100, while values below 70 are usually considered weak. Values below 50 aren't shown; in fact, branches below 50 are collapsed and shown as a polytomy.

Worth repeating.
The branches are collapsed because they are not supported. The analysis cannot tell the relationship between the branches. For example the cladistic analysis cannot determine if creatures like oviraptorids are ancestral or descendant to basal Paraves.
Authors continually fail to understand this basic point and put out unsupported cladograms.
If researchers and authors acknowledged this point they would then have to assess the evidence about where to place oviraptorids. The cladistic analysis does not help in that regard.

For example, the fact that the oviraptorids are dated tens of millions of years later (closer to today) than the basal Paraves would actually settle the issue.

I will be leaving this topic soon. I know you folks need advance warning.

It is important to realize that the ad hoc hypotheses of the ghost lineages and the exaptations are only required because of the idea that creatures like oviraptorids were ancestral to basal Paraves. But that placement is not supported by the cladistic analyses.

People here are at the "pretending not to understand the subject and/or pretending not to understand that there is a problem" stage. This is the first stage. You have not even reached the "denial" stage. From experience we can know that the only way out of this first stage is for someone brave enough to admit seeing that there is a problem. Then the "denial" stage can begin. Followed by the "yes but" stage.
  • Last Edit: June 19, 2017, 08:20:19 AM by socrates1

Re: Bootstrapping
Reply #112
Socrates is obviously a polygamist.

  • socrates1
Re: Bootstrapping
Reply #113
http://ib.berkeley.edu/courses/ib200a/labs/ib200a_lab10_bootstrap_jackknife_bremer.pdf
Bootstrapping calculates a support value for each node based on the fraction of samples that support that node. The highest support value is 100, while values below 70 are usually considered weak. Values below 50 aren't shown; in fact, branches below 50 are collapsed and shown as a polytomy.

Worth repeating.
The branches are collapsed because they are not supported. The analysis cannot tell the relationship between the branches. For example the cladistic analysis cannot determine if creatures like oviraptorids are ancestral or descendant to basal Paraves.
Authors continually fail to understand this basic point and put out unsupported cladograms.
If researchers and authors acknowledged this point they would then have to assess the evidence about where to place oviraptorids. The cladistic analysis does not help in that regard.

For example, the fact that the oviraptorids are dated tens of millions of years later (closer to today) than the basal Paraves would actually settle the issue.

I will be leaving this topic soon. I know you folks need advance warning.

It is important to realize that the ad hoc hypotheses of the ghost lineages and the exaptations are only required because of the idea that creatures like oviraptorids were ancestral to basal Paraves. But that placement is not supported by the cladistic analyses.

People here are at the "pretending not to understand the subject and/or pretending not to understand that there is a problem" stage. This is the first stage. You have not even reached the "denial" stage. From experience we can know that the only way out of this first stage is for someone brave enough to admit seeing that there is a problem. Then the "denial" stage can begin. Followed by the "yes but" stage.

The funny thing is that the huge polytomy is staring everyone in the face in the Supplementary Information sections of these studies. It is very odd that people can look at that and pretend that is not a problem. But of course the first stage is not a failure of perception. It is willful blindness. So anything can be ignored. Even the authors of the studies can overlook the evidence that they themselves have published.

  • socrates1
Re: Bootstrapping
Reply #114
http://ib.berkeley.edu/courses/ib200a/labs/ib200a_lab10_bootstrap_jackknife_bremer.pdf
Bootstrapping calculates a support value for each node based on the fraction of samples that support that node. The highest support value is 100, while values below 70 are usually considered weak. Values below 50 aren't shown; in fact, branches below 50 are collapsed and shown as a polytomy.

Worth repeating.
The branches are collapsed because they are not supported. The analysis cannot tell the relationship between the branches. For example the cladistic analysis cannot determine if creatures like oviraptorids are ancestral or descendant to basal Paraves.
Authors continually fail to understand this basic point and put out unsupported cladograms.
If researchers and authors acknowledged this point they would then have to assess the evidence about where to place oviraptorids. The cladistic analysis does not help in that regard.

For example, the fact that the oviraptorids are dated tens of millions of years later (closer to today) than the basal Paraves would actually settle the issue.

I will be leaving this topic soon. I know you folks need advance warning.

It is important to realize that the ad hoc hypotheses of the ghost lineages and the exaptations are only required because of the idea that creatures like oviraptorids were ancestral to basal Paraves. But that placement is not supported by the cladistic analyses.

People here are at the "pretending not to understand the subject and/or pretending not to understand that there is a problem" stage. This is the first stage. You have not even reached the "denial" stage. From experience we can know that the only way out of this first stage is for someone brave enough to admit seeing that there is a problem. Then the "denial" stage can begin. Followed by the "yes but" stage.

The funny thing is that the huge polytomy is staring everyone in the face in the Supplementary Information sections of these studies. It is very odd that people can look at that and pretend that is not a problem. But of course the first stage is not a failure of perception. It is willful blindness. So anything can be ignored. Even the authors of the studies can overlook the evidence that they themselves have published.

It is funny that after decades, the best that cladistics and the calculation of support values can show is this huge polytomy (Euparaves). It is an acknowledgement that they do not actually know anything.
Thank goodness for the support values (bootstrap/jackknife) or they could have kept bluffing forever. I am not coming up with new controversial evidence. I am simply reporting what they have documented. 
  • Last Edit: June 19, 2017, 08:44:49 AM by socrates1

Re: Bootstrapping
Reply #115
Its funny that there is absolutely nothing linking birds and pterosaurs to each other, and a whole heap of things that link coelurosaurians together.

And yet Sucky cannot acknowledge it even if it is staring him in the face.
Why do I bother?

  • socrates1
Re: Bootstrapping
Reply #116
Quote
The funny thing is that the huge polytomy is staring everyone in the face in the Supplementary Information sections of these studies. It is very odd that people can look at that and pretend that is not a problem. But of course the first stage is not a failure of perception. It is willful blindness. So anything can be ignored. Even the authors of the studies can overlook the evidence that they themselves have published.

It is funny that after decades, the best that cladistics and the calculation of support values can show is this huge polytomy (Euparaves). It is an acknowledgement that they do not actually know anything.
Thank goodness for the support values (bootstrap/jackknife) or they could have kept bluffing forever. I am not coming up with new controversial evidence. I am simply reporting what they have documented. 

For those who are actually interested in understanding this topic more precisely look at Figure 1 here:
http://www.cell.com/cms/attachment/2019474733/2039573089/gr1.jpg


Notice they do not use the huge polytomy. But also notice from the text that they rely on the nodes that should have been collapsed to make their case.

If you are really attentive, you will have noticed that even in Figure S1 and S2 (in the Supplementary Information section) that they do not post the huge polytomy. But it is there, they just find a way to not actually post it.
http://www.cell.com/cms/attachment/2019474733/2039573094/mmc1.pdf
  • Last Edit: June 19, 2017, 09:20:59 AM by socrates1

Re: Bootstrapping
Reply #117
Hey sucky, did you read this in the abstract?

Quote
Birds evolved significantly faster than other theropods, but they are indistinguishable from their closest relatives in morphospace. Our results demonstrate that the rise of birds was a complex process: birds are a continuum of millions of years of theropod evolution, and there was no great jump between nonbirds and birds in morphospace, but once the avian body plan was gradually assembled, birds experienced an early burst of rapid anatomical evolution.
You're straight up lying when you say there's "no connection" between tyrannosaurs and birds.
Why do I bother?

  • socrates1
Re: Bootstrapping
Reply #118
Quote
The funny thing is that the huge polytomy is staring everyone in the face in the Supplementary Information sections of these studies. It is very odd that people can look at that and pretend that is not a problem. But of course the first stage is not a failure of perception. It is willful blindness. So anything can be ignored. Even the authors of the studies can overlook the evidence that they themselves have published.

It is funny that after decades, the best that cladistics and the calculation of support values can show is this huge polytomy (Euparaves). It is an acknowledgement that they do not actually know anything.
Thank goodness for the support values (bootstrap/jackknife) or they could have kept bluffing forever. I am not coming up with new controversial evidence. I am simply reporting what they have documented. 

For those who are actually interested in understanding this topic more precisely look at Figure 1 here:
http://www.cell.com/cms/attachment/2019474733/2039573089/gr1.jpg


Notice they do not use the huge polytomy. But also notice from the text that they rely on the nodes that should have been collapsed to make their case.

If you are really attentive, you will have noticed that even in Figure S1 and S2 (in the Supplementary Information section) that they do not post the huge polytomy. But it is there, they just find a way to not actually post it.
http://www.cell.com/cms/attachment/2019474733/2039573094/mmc1.pdf

Does anyone notice from the text that they rely on the nodes that should have been collapsed to make their case? Their entire case is based on nodes that actually should have been collapsed because they are not supported.
  • Last Edit: June 19, 2017, 09:36:02 AM by socrates1

Re: Bootstrapping
Reply #119
When their case is:

Quote
Tyrannosauroids are the most basal major coelurosaurian subgroup; therizinosauroids and alvarezsauroids form a clade with oviraptorosaurs and paravians exclusive of more basal coelurosaurs; therizinosauroids and oviraptorosaurs are not sister taxa; and, for the first time, a TWiG analysis recovers a polytomy between avialans, dromaeosaurids, and troodontids, meaning that the immediate relative of birds cannot be clearly identified.

Then no, they do not, you moronic fuck.
Why do I bother?

  • socrates1
Re: Bootstrapping
Reply #120
Quote
The funny thing is that the huge polytomy is staring everyone in the face in the Supplementary Information sections of these studies. It is very odd that people can look at that and pretend that is not a problem. But of course the first stage is not a failure of perception. It is willful blindness. So anything can be ignored. Even the authors of the studies can overlook the evidence that they themselves have published.

It is funny that after decades, the best that cladistics and the calculation of support values can show is this huge polytomy (Euparaves). It is an acknowledgement that they do not actually know anything.
Thank goodness for the support values (bootstrap/jackknife) or they could have kept bluffing forever. I am not coming up with new controversial evidence. I am simply reporting what they have documented. 

For those who are actually interested in understanding this topic more precisely look at Figure 1 here:
http://www.cell.com/cms/attachment/2019474733/2039573089/gr1.jpg


Notice they do not use the huge polytomy. But also notice from the text that they rely on the nodes that should have been collapsed to make their case.

If you are really attentive, you will have noticed that even in Figure S1 and S2 (in the Supplementary Information section) that they do not post the huge polytomy. But it is there, they just find a way to not actually post it.
http://www.cell.com/cms/attachment/2019474733/2039573094/mmc1.pdf

Does anyone notice from the text that they rely on the nodes that should have been collapsed to make their case? Their entire case is based on nodes that actually should have been collapsed because they are not supported.

It is tedious dealing with people who pretend that they cannot understand anything. Is there anyone who actually understands the really quite straightforward point I am making? Anyone brave enough to get beyond Stage 1? (In the past I have said that Stage 1 is the most tedious stage.)

  • socrates1
Re: Bootstrapping
Reply #121
Quote
The funny thing is that the huge polytomy is staring everyone in the face in the Supplementary Information sections of these studies. It is very odd that people can look at that and pretend that is not a problem. But of course the first stage is not a failure of perception. It is willful blindness. So anything can be ignored. Even the authors of the studies can overlook the evidence that they themselves have published.

It is funny that after decades, the best that cladistics and the calculation of support values can show is this huge polytomy (Euparaves). It is an acknowledgement that they do not actually know anything.
Thank goodness for the support values (bootstrap/jackknife) or they could have kept bluffing forever. I am not coming up with new controversial evidence. I am simply reporting what they have documented. 

For those who are actually interested in understanding this topic more precisely look at Figure 1 here:
http://www.cell.com/cms/attachment/2019474733/2039573089/gr1.jpg


Notice they do not use the huge polytomy. But also notice from the text that they rely on the nodes that should have been collapsed to make their case.

If you are really attentive, you will have noticed that even in Figure S1 and S2 (in the Supplementary Information section) that they do not post the huge polytomy. But it is there, they just find a way to not actually post it.
http://www.cell.com/cms/attachment/2019474733/2039573094/mmc1.pdf

Does anyone notice from the text that they rely on the nodes that should have been collapsed to make their case? Their entire case is based on nodes that actually should have been collapsed because they are not supported.

It is tedious dealing with people who pretend that they cannot understand anything. Is there anyone who actually understands the really quite straightforward point I am making? Anyone brave enough to get beyond Stage 1? (In the past I have said that Stage 1 is the most tedious stage.)


From the study:
http://www.cell.com/current-biology/fulltext/S0960-9822(14)01047-1
"therizinosauroids and alvarezsauroids form a clade with oviraptorosaurs and paravians exclusive of more basal coelurosaurs"

But they do not use that conclusion in their case. That polytomy is because of the unsupported nodes.

Even that is not complete because they have not shown that "Maniraptora" is also a node that should be collapsed.
  • Last Edit: June 19, 2017, 10:12:37 AM by socrates1

  • VoxRat
  • wtactualf
Re: Bootstrapping
Reply #122
Oh lord...

"Socrates" is back to his "stages".

::)
"I understand Donald Trump better than many people because I really am a lot like him." - Dave Hawkins

Re: Bootstrapping
Reply #123
anyone trying to use bootstrap/jackknife as a measure of clade support is wrong. that's not what it means. hth.

Thanks, that's good to know.  What statistical metric(s) would be appropriate to use as a measure of clade support?

realistically you need to make a constraint tree and then test it against random trees using I dunno a KH or something. Node support is ridiculously hard to calculate in a meaningful way and most of the methods used to generate node support do not actually do this.

Bootstrapping and Jackknifing basically estimate how well the universe of characters that inform a specific node have been sampled. They are notoriously easy to calculate and notoriously difficult to interpret. It definitely doesn't reflect how much you should trust a node per se and doesn't tell you anything about broader trends in treespace, though.

Basically trying to treat any of these numbers as having mathematical or biological meaning is difficult to justify though.

  • socrates1
Re: Bootstrapping
Reply #124
Quote
The funny thing is that the huge polytomy is staring everyone in the face in the Supplementary Information sections of these studies. It is very odd that people can look at that and pretend that is not a problem. But of course the first stage is not a failure of perception. It is willful blindness. So anything can be ignored. Even the authors of the studies can overlook the evidence that they themselves have published.

It is funny that after decades, the best that cladistics and the calculation of support values can show is this huge polytomy (Euparaves). It is an acknowledgement that they do not actually know anything.
Thank goodness for the support values (bootstrap/jackknife) or they could have kept bluffing forever. I am not coming up with new controversial evidence. I am simply reporting what they have documented. 

For those who are actually interested in understanding this topic more precisely look at Figure 1 here:
http://www.cell.com/cms/attachment/2019474733/2039573089/gr1.jpg


Notice they do not use the huge polytomy. But also notice from the text that they rely on the nodes that should have been collapsed to make their case.

If you are really attentive, you will have noticed that even in Figure S1 and S2 (in the Supplementary Information section) that they do not post the huge polytomy. But it is there, they just find a way to not actually post it.
http://www.cell.com/cms/attachment/2019474733/2039573094/mmc1.pdf

Does anyone notice from the text that they rely on the nodes that should have been collapsed to make their case? Their entire case is based on nodes that actually should have been collapsed because they are not supported.

It is tedious dealing with people who pretend that they cannot understand anything. Is there anyone who actually understands the really quite straightforward point I am making? Anyone brave enough to get beyond Stage 1? (In the past I have said that Stage 1 is the most tedious stage.)


From the study:
http://www.cell.com/current-biology/fulltext/S0960-9822(14)01047-1
"therizinosauroids and alvarezsauroids form a clade with oviraptorosaurs and paravians exclusive of more basal coelurosaurs"

But they do not use that conclusion in their case. That polytomy is because of the unsupported nodes.

Even that is not complete because they have not shown that "Maniraptora" is also a node that should be collapsed.

One very clever way to deal with the problem of the huge polytomy is to say that support values are meaningless anyway*. The problem with that is that none of the authors of the studies make anything like that point. They go to the trouble of calculating the support values and then document the results in the Supplementary Information.
The problem is not with support values. The problem is that the authors calculate the support values and then ignore them as if they had never been calculated.
Does anyone understand that straightforward point?

* For those who are really with it, you will realize that that is an example of the "So what" stage which is a much later stage.
  • Last Edit: June 19, 2017, 11:10:53 AM by socrates1