On 6/29/12 5:10 AM, in article
e7229162-dc55-4e74...@googlegroups.com, "Onion Knight"
<
onionkn...@hotmail.com> wrote:
...
>>> What is a lie?
>>
>> No wonder you so rarely tell the truth: you do not even know that a lie is
>> something you knowingly tell which is contrary to the truth.
>>
>
> I think you're confused. Obviously from the context he was asking which of his
> statements were a lie. Do you have trouble with context? It seems like you
> totally misunderstand most things people write. Even in this thread you took
> statements CC made where he was making fun of you, and tried to pass them off
> as if they were legitimate statements. If it's a deliberate troll technique,
> then I think it works! If it's not, then I think you need to try harder or get
> some help, because responses like your's above only make you look really,
> really, dumb.
Given cc's history, it is reasonable to assume he cannot differentiate
between the concepts as to what is a lie and what is not. But, as you know,
I have pointed to many of his lies.
As far as taking statements made by cc where he was making fun of me and
then trying to pass them off as something else, that is one of his lies. I
did no such thing. He repeatedly admitted that there was a drop in the data
in 2012. When called on it he tried to retcon his claims.
>>> Did the rest of the world see an increase in Linux usage? Nope.
>>
>> This is not in contention.
>>
>>> Did you originally state your prediction was only for the last part of 2011?
>>> Nope.
>>
>> This is not in contention.
>>
>>> So if you're saying there's a correlation, then it's only for the last
>>> part of 2011 and only for California.
>>
>> See: you are clueless on what you have been told. And you snip and run from
>> your mistakes:
>
> But if the above two are not in contention, then it's true that the
> correlation was only for the latter part of 2011 and only for California like
> CC says. That doesn't seem like much of a correlation, and also doesn't fit
> your original prediction at all. So it seems like your prediction was totally
> wrong. What am I missing?
The correlation fit the full data set. Now we know that this came from some
data in CA. This does not change the fact that the correlation fit the
whole data set.
What it does do is very strongly weaken my claim of cause and effect.
Correlations and cause-and-effects are not the same. The only ones who
confuse these concepts as you just did are cc and Carroll - and thus I
suspect your posts are being forged by one of them. Given the style of the
writing I would guess Carroll, but I have no proof.
>> 1) You were wrong to claim it was a "fact" that some data points
>> were outliers: the reality is that there is no clear
>> definition nor single standard way to determine such, and thus
>> such determinations are subjective and *not* "facts".
>
> I found plenty of calculators online to find outliers. All of them said the
> data had outliers in them, and all of them had pretty much the same outliers.
> What way to find outliers did you use came up with no outliers as a result?
Some calculators find some. Some do not. But since the distribution is not
even appropriate for most of these calculators, and since different
calculators disagree on which points, if any, are outliers, then his claim
that it is a "fact" that some set of specific points are outliers is
completely absurd. It has been proved wrong.
>> 2) You made a rather silly claim to say that 17% of the data are
>> "outliers" - especially when those data are in a direct series
>> tied to a *very* clear trend that includes 25% of the data.
>
> Can outliers not be in a direct series? I thought outliers were random
> occurances, or at least not predictable. You can flip a coin a thousand times
> and have heads come up. That doesn't mean it's a pattern. I didn't see
> anything online where there had to be only a certain percentage of outliers,
> and no more. Also, I think 17% is the wrong figure, looking at the data you
> put up from 2007. It's more like 7% I think.
If outliers are a high percentage of the data then the data is likely too
corrupt to be of value. In any case, the 17% is in reference to the original
data set cc asked me to use, not the one he claims to have run to when he
realized how absurd his claims were (the data set he was never able to
produce).
>> 3) Outliers are atypical - you clearly are stretching to call 17%
>> of the data, which occurs in a clear pattern, "atypical" to the
>> data. Tied to the top 3: <
http://tinyurl.com/7hymgfg>. Makes
>> it very clear just how clueless cc is.
>
> I think you're seeing a pattern where none exists. See my comment about the
> coin flip above. Again, I really doubt it is 17%, and I believe cc posted in
> another thread it was more like 6%, which I also think is wrong. It was 7-8% I
> believe depending on which online outlier calculator you used.
First, some of cc's claims on this issue:
cc:
-----
There are four outliers in the latter half of 2011.
-----
There are some low numbers sprinkled in there from late 2007,
early 2008, that are also outliers, if you can't figure that
out for yourself.
-----
But that doesn't change the fact that those data points are
statistically insignificant because they are outliers.
-----
There is a well-defined (several actually) method for
determining outliers. This is a fact. I found the outliers,
you did not.
-----
Your new trendline uses data made up of mostly outliers,
evidenced by all standard measurements of outliers and the
fact that a single state caused the fluctuation.
-----
Now looking at different outlier calculators:
<
http://tmp.gallopinginsanity.com/median_and_array.pdf>
It is 0% using the Median() and array formula from:
<
http://tinyurl.com/7fqxztv>. This is true at when you use a Z-Score
Tolerance of 3.5 (the norm), and also all the way down to 2. Even at a
tolerance of 1.5 there is only one outlier that is detected, and that is in
Jan of 2012, not in the "latter half of 2011" as cc claimed... and certainly
not four from the latter half... and *none* from "late 2007, early 2008".
At a tolerance of 1 (and here we are getting far from the norm of 3.5!), you
still only have four points which are found to be outliers: 2 (not 4) from
the "latter half of 2011" and only one from "late 2007" and none from "early
2008". By the way, if you want the Excel spreadsheet I used I have made it
available: <
http://tmp.gallopinginsanity.com/median_and_array.xls>. Unlike
cc I am very open with how I calculate these things. He is not because he
just makes things up.
So let us look at Grubbs: <
http://tinyurl.com/832dvjz>:
Only one outlier, and it is at the start of 2012, not in the "latter half of
2011" as cc claimed... and certainly not four from the latter half... and
*none* from "late 2007, early 2008".
On the same link we have the Quartile method: and it finds 3 outliers. Two
are from the "latter half of 2011" (not four, as cc claimed), and none are
from "late 2007, early 2008"... and one is from 2012 - a time period cc
never mentioned in his claims about where he found outliers.
And yet cc went on and on about how he found these - but he never has been
specific as to his method and none of the calculators nor methods I have
tried have replicated his claimed findings... findings he cannot show.
Worse for him, he behaves like his findings are a foregone conclusion - and
that has been completely trounced. He is flat out wrong. He is completely
clueless on the topic.
>> 4) You have clearly placed a great deal of focus on the
>> correlation coefficient even though it is unwise to do so at
>> the expense of a focus on the visual graphs / visible trends.
>
> Your eyes can lie. Why wouldn't an exact calculation be important?
Who said it would not be? But what "exact calculation"? cc presented
*none*... and the data itself does not approximate a normal distribution and
is not linear... this the calculations we are discussing are of limited
value anyway.
> All of the correlation results I find online have a coefficient. Is there are
> a reason you shouldn't calculate a correlation coefficient? I'm confused, I
> thought that was standard procedure.
cc insists on including the calculation of outliers - which, as you have
been shown, is not a "standard" operation with one accepted method and one
accepted result. It is not a "fact" as he claimed.
The linear trend line itself is a poor tool to use, given the non-linear
nature of the data.
> Why would someone place a greater deal of focus on the inexact and potentially
> wrong method of looking at a graph, instead of getting a numerical result? You
> think people should place a greater focus on estimation at the expense of
> exact results?
Incorrect. I go into far more detail in one of the links in my sig:
> I've seen your graph of Linux usage, and data set. It's been very helpful
> verifying what CC said.
How so? What claim?
> But I haven't seen your graph correlating Linux usage with an increased focus
> in usability. Do you have that graph? It would be hard to focus on the visual
> graphs/visible trends if you don't even have anything visible to show! Thanks.
What graph? I have been very open that the graph does not show the
cause-and-effect I spoke of, and my prediction was vague... it is not like I
had a predictive graph and can show some numerical calculation which shows
how well it correlated to the actual data.
What I did was made a vague prediction about an increase and gave a reason
for my prediction. The data correlated with that prediction (during the
increase... though frankly that was more of an increase than I expected,
again, my predictions were vague). It has now been shown (by bbgruff) that
the increase was localized to a small region... so not only does the data
from 2012 go against my predictions, the data from the latter half of 2011
does not fit them well, either. This has been established and discussed and
is not in contention.
So given that what graph are you looking for? Your question shows a level
of ignorance displayed only by cc and Carroll - which is yet another reason
to believe your post has been forged by one of them (likely Carroll, but
either one is possible and neither is proved).
>> 5) You were wrong to deny that the sigma lines can be correctly
>> drawn based on the distance from the mean to the inflection
>> points.
>
> Reading over his posts, I don't see where he did that.
cc:
-----
There'se nothing wrong with the image, other than some weird
axis labeling.
-----
Snit's so fucking stupid he thinks the sigma lines are drawn
based on distance from the mean, not area under the curve.
-----
| The sigma lines are drawn based on the area of the curve -
| which is easy to see when the images screw it up, esp. when
| they do so really badly, like in some of the ones I showed
| you.
They are not wrong.
------
LOL!!!! All of those links are fine. The first sigma lines
cover 68% of the area UNDER THE CURVE.
-----
If you would like to prove, on any single one of the links
you call incorrect, that the first sigma lines do not bound
an area that is 68.2% of the area UNDER THE CURVE, then I
would like to see it.
-----
Hahahaha your "approximate inflection points" are hilarious.
Please, post more on this subject.
------
I know exactly what an inflection point is. It's where the
second derivative changes sign, and it's exactly where the
sigma lines are in your supposed incorrect examples. Funny
how you're now questioning the applications used to generate
those graphs! Face it, you're wrong.
------
His denials were repeated and strong. He insisted it was based on the
percentage under the curve and *not* the distance from the mean to
*anything*.
> He said there is no exact numerical distance, and he also said they are at the
> inflection points. So again, I wonder if you have some sort of problem
> reading? I think you're confused on this. Look at the links you posted, it's
> obvious that many are not normal distributions, so the inflection point isn't
> meaningful.
Look at his quotes. Also, I had already discussed how the reference to
distance was to some feature of the curve and not in inches or centimeters
or the like. He kept making his claims. He was completely ignorant of this
- and he knows it, hence the reason he repeatedly runs.
>> 6) You were wrong to deny that the poorly done depictions I showed
>> you were, in fact, poorly done. They were unambiguously wrong.
>
> You said all the links you posted were examples of incorrect placement of
> standard deviation lines. CC claims many were done with MatLab. Looking
> online, that looks to be the case. So how could they be poorly done if they
> were in MatLab? Why is MatLab poorly done software?
Please show the examples I stated were bad that were not. So far cc has
completely failed to find a single example.
But here is the rub: say he does find one. It is *possible* I was mistaken
and I would admit to it if it were to be shown. Unlike cc I am honest.
Now look at cc's claims above. His first:
cc:
-----
There'se nothing wrong with the image, other than some weird
axis labeling.
-----
Was in reference to this image:
<
http://www.robertniles.com/stats/graphics/stndv.gif>
Once you know to look for the inflection points it is *clearly* an incorrect
depiction. There is no doubt cc was wrong. He will, however, never admit
to it. He simply runs.
>> 7) You were wrong to say I missed steps in the creation of a
>> linear trend line in Excel. I did no such thing (and you
>> never were able to list any steps I missed in creating a
>> linear trend line, nor explain why the MS site and others
>> would also miss these "steps" you spoke about)
>
> I read his comments, and he said he followed MS directions. So you're lying
> here. Why lie? You admitted you never checked for outliers, and let them screw
> up your data. So perhaps that's your missing step?
If you think I missed steps in creating a linear trend line in MS's
directions then please state what steps: <
http://tinyurl.com/25x2dvn>
I did use the right-click method of getting to the tools and not the Ribbon
- but that is not a "missing step" it is merely using a different method to
get to the same tools.
The claim I missed steps is an out and out lie. It has been completely
disproved.
>> 8) You were wrong to say I was pushing the correlations I noted
>> as being proof of the causation I had spoken of earlier. I
>> did no such thing.
>
> In the thread you reference, he admits he was wrong to say you were predicting
> a causation.
I do not recall that. Can you link to it or quote his admission there? If
he did so I will happily accept his admission and note this point has been
"handled" by him.
> So yes, he was wrong and he admits it. What is the problem with
> that? I don't see you admitting you were wrong about your prediction.
Oh, if he made such an admission I would love to see it. I would love to be
proved wrong.
As far as my admissions about errors, see above and many other posts. I do
not hide nor deny my errors.
>> 9) You repeatedly snip and ignore comments which are contrary to
>> your claimed views.
>
> I already found many spots where he already answered these questions.
No. Given that he has not you have not found such. If you think otherwise
then quote or link. An empty claim is meaningless... and that is all you
have there. Very much like cc and Carroll. Gee, not saying your ID is
being forged, just noting all data points to that conclusion. :)
>> These things are *not* in contention either. Hence why you snip and run.
>>
>>> That's pretty far from your original statement.
>>
>> One: your lies are very different than my statements. That is not in
>> contention.
>>
>> Two: as we get new data I alter my views... as is the rational and honest
>> thing to do. This is what you are incapable of doing... to the point that
>> you see it as a bad thing when I do it. Yes: you think it is *bad* that I
>> alter my views based on new data.
>
> But your views were wrong at the time, with no new data!
I am open to you supporting this.
> And there was no new data when CC posted the info claiming you were wrong.
He makes many idiotic claims. So?
> You were saying you were right, up until yesterday. So when you were told you
> were wrong with old data, you said you were right. When you were shown you
> were wrong with new data, you still said you were right. In another post you
> said you were willing to admit when you are wrong. Well will you finally admit
> it? I doubt it.
Can you repeat that in English? Thanks! If you think I was wrong about
something then say what! If you can show I was wrong, or if I can, I will
admit to it. It is what I do.
I am not like cc and Carroll.
>>> But hey, if your tiny brain in your fat head somehow sees that as being
>>> "right" then what can I do?
>>
>> You do realize that you have made it clear you know you are lying, right?
>>
>> I keep asking you that and you keep running. You hate knowing how obvious
>> your lies are.
>
> I found where CC answers all the questions you said he was running from.
Given how he has not this is not possible. But feel free to try to show
otherwise.
> I verified all the answers he gave as best I could. In the end, his statements
> match up perfectly to the real world results, and show the prediction you made
> was incorrect. So what is he lying about?
See above: you and he offer no verification for his claims. None. I show
massive amounts of counters to his claims.
> Was there an increase in Linux usage anywhere in the world? There wasn't.
Well, in CA there was, or at least that is what the data shows.
> Did CC's trendline show an increase in Linux usage? No, it didn't. Did your
> trendline show an increase in Linux usage? Yes it did.
And here is the overall trend: <
http://tinyurl.com/832dvjz>
Now we can remove some data points from the early and late dates - the ones
which some method show as outliers, but the trend line will still be clearly
an upward one. Of course, as I have noted, this is not linear data so the
predictive value, esp. in the short term, is pretty much useless. Over the
long run, though, it does suggest the trend will continue upward.
Also keep in mind that the later data
> Did California skew the results? Yes, it did. Did CC correctly identify which
> points were skewed using outliers?
Absolutely not.
> Yes, he did.
I would love to see your support for this.
> Did correctly identify which points were skewed using outliers? No, you
> didn't.
The search for outliers is not what showed the increase applied only to CA.
> You even claimed there were no outliers at all.
Please quote this claim. I have noted that the number of points which are
outliers, if any, is largely subjective and can be based on any number of
methods. I have shown where at least some of these methods show no
outliers, but others do show some (though none that I find match what cc
claimed to find - and he has not been able to produce, say, an Excel
spreadsheet he claimed he used to calculate these things). There is no
reason to think such a spreadsheet even exists... but if it does he is
welcome to email it to me and I will post it for him. He has emailed me in
the past and I have no problem with him doing this so he can support his
claims. He can then look at the link and make sure I have not modified it
in any way (he can even password protect it if he wishes).
> So it seems like you're just upset that CC has been right this entire time and
> his graphs and analysis matched up perfectly with real-world results. Why are
> you so upset that CC is right and you are wrong?
Yawn. Not even you, cc or Carroll, believe this BS you are spewing... but
it was fun to read your forgery and refute your claims. :)
On cc's claims about Grubbs and Quartiles: <
http://tinyurl.com/75yvo8z>