You can ahahaha all you want. Linear is bullshit, quadratic is bullsht,
exponential is bullshit etc. The only model which wouldn't be bullshit
would be an INFORMED model that takes into account physics. This model
has yet to be built.

On Sun, 11 May 2008 01:38:57 -0400, V-for-Vendicar wrote:
"zdzis1" <zdz... at (no spam) 31.pl> wrote
Now, feel free to compute the statistical probability that the slope
is positive, and negative.

Here you are:
the 95% confidence interval for the slope is (see the R output above)
(0.018121-1.96*0.009951, 0.018121+1.96*0.009951)=(-0.00138296,
0.03762496) thus it could be positive and it could be negative - that's
all we know.

I'm sorry. That didn't answer the question put to you...

What is the probability that the slope is positive? And what is the
probability that the slope is negative?

You were asked for two probabilities, not a single combined probability.

MMMMMMMMMMOOOOOOOOOORRRRRRRRRRRROOOOOOOONNNNNNNNNN

I gave you a typical answer you get in statistics. But if you insist on
separating the two probabilities then OK. Nothing could be easier

slope/sd_slope (where sd_slope is the standard deviation of the slope)
has a t distribution on n-2 degrees of freedom if H0 holds (you were told
to check what H0 is here, so I assume you have done your homework). Now
integrating this t distribution from -inf to -1.82 (which is
just-0.018121/0.009951) we get

1) probability that the slope is negative: 0.053
2) probability that the slope is positive: 0.947

Now, go to the library and check what this means in statistical terms.

"zdzis1" <zdz... at (no spam) 31.pl> wrote
the linear model is obviously bullshit here,

Oh, I see, it should be quadratic or something. Ahahahahahahahahah

You can ahahaha all you want. Linear is bullshit, quadratic is bullsht,
exponential is bullshit etc. The only model which wouldn't be bullshit
would be an INFORMED model that takes into account physics. This model
has yet to be built.



MMMMMMMMOOOOOOOOOOORRRRRRRRRRRROOOOOOOOONNNNNNNNNN

Now here is your opportunity to answer the question put to you.

I have. Now, I remembered asking you a question: what does p-value > 0.05
mean for linear regression. And more speciffically, does it or does it
not mean that the slope could be negative in the discussed case? You are
in a better situation than myself, because you can find your answer in a
book - I have to know mathematics to actually calculate the answers to
your dumb questions.

On May 11, 2:50 pm, zdzis1 <zdz... at (no spam) 31.pl> wrote:





On Sun, 11 May 2008 01:38:57 -0400, V-for-Vendicar wrote:
"zdzis1" <zdz... at (no spam) 31.pl> wrote
  Now, feel free to compute the statistical probability that the slope
  is positive, and negative.

Here you are:
the 95% confidence interval for the slope is (see the R output above)
(0.018121-1.96*0.009951, 0.018121+1.96*0.009951)=(-0.00138296,
0.03762496) thus it could be positive and it could be negative - that's
all we know.

I'm sorry.  That didn't answer the question put to you...

What is the probability that the slope is positive? And what is the
probability that the slope is negative?

You were asked for two probabilities, not a single combined probability.

  MMMMMMMMMMOOOOOOOOOORRRRRRRRRRRROOOOOOOONNNNNNNNNN

I gave you a typical answer you get in statistics. But if you insist on
separating the two probabilities then OK. Nothing could be easier

slope/sd_slope (where sd_slope is the standard deviation of the slope)
has a t distribution on n-2 degrees of freedom if H0 holds (you were told
to check what H0 is here, so I assume you have done your homework). Now
integrating this t distribution from -inf to -1.82 (which is
just-0.018121/0.009951) we get

1) probability that the slope is negative: 0.053
2) probability that the slope is positive: 0.947

Now, go to the library and check what this means in statistical terms.

"zdzis1" <zdz... at (no spam) 31.pl> wrote
the linear model is obviously bullshit here,

  Oh, I see, it should be quadratic or something.  Ahahahahahahahahah

You can ahahaha all you want. Linear is bullshit, quadratic is bullsht,
exponential is bullshit etc. The only model which wouldn't be bullshit
would be an INFORMED model that takes into account physics. This model
has yet to be built.

   MMMMMMMMOOOOOOOOOOORRRRRRRRRRRROOOOOOOOONNNNNNNNNN

    Now here is your opportunity to answer the question put to you..

I have. Now, I remembered asking you a question: what does p-value > 0.05
mean for linear regression. And more speciffically, does it or does it
not mean that the slope could be negative in the discussed case? You are
in a better situation than myself, because you can find your answer in a
book - I have to know mathematics to actually calculate the answers to
your dumb questions.

Nope. If you knew the mathematical statistics behind data anlysis you
wouldn't be messing with regression on single slices of the data. The
best buy procedure is to take periodic trending samples over the whole
range of available years.

- Show quoted text -

On May 11, 2:50 pm, zdzis1 <zdz... at (no spam) 31.pl> wrote:
On Sun, 11 May 2008 01:38:57 -0400, V-for-Vendicar wrote:
"zdzis1" <zdz... at (no spam) 31.pl> wrote
Now, feel free to compute the statistical probability that the
slope is positive, and negative.

Here you are:
the 95% confidence interval for the slope is (see the R output
above) (0.018121-1.96*0.009951,
0.018121+1.96*0.009951)=(-0.00138296, 0.03762496) thus it could be
positive and it could be negative - that's all we know.

I'm sorry. That didn't answer the question put to you...

What is the probability that the slope is positive? And what is the
probability that the slope is negative?

You were asked for two probabilities, not a single combined
probability.

MMMMMMMMMMOOOOOOOOOORRRRRRRRRRRROOOOOOOONNNNNNNNNN

I gave you a typical answer you get in statistics. But if you insist on
separating the two probabilities then OK. Nothing could be easier

slope/sd_slope (where sd_slope is the standard deviation of the slope)
has a t distribution on n-2 degrees of freedom if H0 holds (you were
told to check what H0 is here, so I assume you have done your
homework). Now integrating this t distribution from -inf to -1.82
(which is just-0.018121/0.009951) we get

1) probability that the slope is negative: 0.053 2) probability that
the slope is positive: 0.947

Now, go to the library and check what this means in statistical terms.

"zdzis1" <zdz... at (no spam) 31.pl> wrote
the linear model is obviously bullshit here,

Oh, I see, it should be quadratic or something. Ahahahahahahahahah

You can ahahaha all you want. Linear is bullshit, quadratic is bullsht,
exponential is bullshit etc. The only model which wouldn't be bullshit
would be an INFORMED model that takes into account physics. This model
has yet to be built.



MMMMMMMMOOOOOOOOOOORRRRRRRRRRRROOOOOOOOONNNNNNNNNN

Now here is your opportunity to answer the question put to you.

I have. Now, I remembered asking you a question: what does p-value
0.05 mean for linear regression. And more speciffically, does it or
does it not mean that the slope could be negative in the discussed
case? You are in a better situation than myself, because you can find
your answer in a book - I have to know mathematics to actually
calculate the answers to your dumb questions.

Nope. If you knew the mathematical statistics behind data anlysis you
wouldn't be messing with regression on single slices of the data. The
best buy procedure is to take periodic trending samples over the whole
range of available years.

On May 11, 7:50 am, zdzis1 <zdz... at (no spam) 31.pl> wrote:

You can ahahaha all you want. Linear is bullshit, quadratic is bullsht,
exponential is bullshit etc. The only model which wouldn't be bullshit
would be an INFORMED model that takes into account physics. This model
has yet to be built.


Cuz you sez so, on the usenet, right?

On Sun, 11 May 2008 08:01:03 -0700, John M. wrote:
On May 11, 2:50 pm, zdzis1 <zdz... at (no spam) 31.pl> wrote:
On Sun, 11 May 2008 01:38:57 -0400, V-for-Vendicar wrote:
"zdzis1" <zdz... at (no spam) 31.pl> wrote
Now, feel free to compute the statistical probability that the
slope is positive, and negative.

Here you are:
the 95% confidence interval for the slope is (see the R output
above) (0.018121-1.96*0.009951,
0.018121+1.96*0.009951)=(-0.00138296, 0.03762496) thus it could be
positive and it could be negative - that's all we know.

I'm sorry. That didn't answer the question put to you...

What is the probability that the slope is positive? And what is the
probability that the slope is negative?

You were asked for two probabilities, not a single combined
probability.

MMMMMMMMMMOOOOOOOOOORRRRRRRRRRRROOOOOOOONNNNNNNNNN

I gave you a typical answer you get in statistics. But if you insist on
separating the two probabilities then OK. Nothing could be easier

slope/sd_slope (where sd_slope is the standard deviation of the slope)
has a t distribution on n-2 degrees of freedom if H0 holds (you were
told to check what H0 is here, so I assume you have done your
homework). Now integrating this t distribution from -inf to -1.82
(which is just-0.018121/0.009951) we get

1) probability that the slope is negative: 0.053 2) probability that
the slope is positive: 0.947

Now, go to the library and check what this means in statistical terms.

"zdzis1" <zdz... at (no spam) 31.pl> wrote
the linear model is obviously bullshit here,

Oh, I see, it should be quadratic or something. Ahahahahahahahahah

You can ahahaha all you want. Linear is bullshit, quadratic is bullsht,
exponential is bullshit etc. The only model which wouldn't be bullshit
would be an INFORMED model that takes into account physics. This model
has yet to be built.

MMMMMMMMOOOOOOOOOOORRRRRRRRRRRROOOOOOOOONNNNNNNNNN

Now here is your opportunity to answer the question put to you.

I have. Now, I remembered asking you a question: what does p-value
0.05 mean for linear regression. And more speciffically, does it or
does it not mean that the slope could be negative in the discussed
case? You are in a better situation than myself, because you can find
your answer in a book - I have to know mathematics to actually
calculate the answers to your dumb questions.

Nope. If you knew the mathematical statistics behind data anlysis you
wouldn't be messing with regression on single slices of the data. The
best buy procedure is to take periodic trending samples over the whole
range of available years.

Didn't I just say that linear regression here is bullshit?

And "taking periodic trending samples" doesn't mean anything here either
- there are many approaches to finding, describing and predicting trends.
You need to first define exactly what "taking periodic trending samples"
means, because it is not a standard name (but use mathematical formalism
to define the term - no analogies or stories - just definitions of your
terms). Then explain why you think that this is the "best buy procedure",
but again - no stories and analogies - use mathematical formalism.

I am always eager to learn and I'll be glad to discuss "taking periodic
trending samples" with you.

Didn't I just say that linear regression here is bullshit?

Ah..yes. But then you proceeded to carry it out and to examine and
discuss the results (although to your credit you stopped short of
checking the residuals and computing leverage). Why was that, when you
could have moved immediately to perform a more appropriate analysis?

The first thing to do when faced with an unfamiliar piece of jargon is
to search for its use elsewhere. The fact that you apparently did not do
so tells me something about you.

So try the web - I got 26 hits on Metacrawler but other search engines
may give more or less than this. Of these, the one that can best meet
your enthusiasm for new knowledge is in a blog by a mathematician:

http://scienceblogs.com/goodmath/2008/05/
selective_data_and_global_warm.php#more

which as luck would have it is on topic for this ng ;-)

Do get back once you've absorbed the general idea. We can discuss modes
of application, and also efficacy, in the context of temperature
records.

Didn't I just say that linear regression here is bullshit?

Ah..yes. But then you proceeded to carry it out and to examine and
discuss the results (although to your credit you stopped short of
checking the residuals and computing leverage). Why was that, when you
could have moved immediately to perform a more appropriate analysis?

The first thing to do when faced with an unfamiliar piece of jargon is
to search for its use elsewhere. The fact that you apparently did not do
so tells me something about you.

So try the web - I got 26 hits on Metacrawler but other search engines
may give more or less than this. Of these, the one that can best meet
your enthusiasm for new knowledge is in a blog by a mathematician:

http://scienceblogs.com/goodmath/2008/05/
selective_data_and_global_warm.php#more

which as luck would have it is on topic for this ng ;-)

Do get back once you've absorbed the general idea. We can discuss modes
of application, and also efficacy, in the context of temperature
records.

On Sun, 11 May 2008 07:33:43 -0700, kT wrote:
On May 11, 7:50 am, zdzis1 <zdz... at (no spam) 31.pl> wrote:

You can ahahaha all you want. Linear is bullshit, quadratic is bullsht,
exponential is bullshit etc. The only model which wouldn't be bullshit
would be an INFORMED model that takes into account physics. This model
has yet to be built.

Cuz you sez so, on the usenet, right?

give me an example of such a model

On May 11, 10:18 am, zdzis1 <zdz... at (no spam) 31.pl> wrote:
On Sun, 11 May 2008 07:33:43 -0700, kT wrote:
On May 11, 7:50 am, zdzis1 <zdz... at (no spam) 31.pl> wrote:

You can ahahaha all you want. Linear is bullshit, quadratic is
bullsht, exponential is bullshit etc. The only model which wouldn't
be bullshit would be an INFORMED model that takes into account
physics. This model has yet to be built.

Cuz you sez so, on the usenet, right?

give me an example of such a model

Here's one, just for starters :

http://www.giss.nasa.gov/research/modeling/gcms.html

Why don't you try learning about them before you go shooting off your
mouth and letting everybody on to how clueless you are. Start here :

http://en.wikipedia.org/wiki/Global_climate_model

On Sun, 11 May 2008 15:01:20 -0700, kT wrote:
On May 11, 10:18 am, zdzis1 <zdz... at (no spam) 31.pl> wrote:
On Sun, 11 May 2008 07:33:43 -0700, kT wrote:
On May 11, 7:50 am, zdzis1 <zdz... at (no spam) 31.pl> wrote:

You can ahahaha all you want. Linear is bullshit, quadratic is
bullsht, exponential is bullshit etc. The only model which wouldn't
be bullshit would be an INFORMED model that takes into account
physics. This model has yet to be built.

Cuz you sez so, on the usenet, right?

give me an example of such a model

Here's one, just for starters :

http://www.giss.nasa.gov/research/modeling/gcms.html

Yawn... it is one of the many models produced by this organisation - none
of which works (this is why they keep changing them). If they did we
would they would have been able to predict last summer or this winter -
which they were not. The fact that people talk about physics while
creating their model does not mean that the model is correct.



Why don't you try learning about them before you go shooting off your
mouth and letting everybody on to how clueless you are. Start here :

http://en.wikipedia.org/wiki/Global_climate_model

please ... don't make me lough

once again - the fact that people create models does not mean that they
are correct

zdzis1 wrote:
On Sun, 11 May 2008 18:09:29 -0700, kT wrote:

On May 11, 5:18 pm, zdzis1 <zdz... at (no spam) 31.pl> wrote:
On Sun, 11 May 2008 15:01:20 -0700, kT wrote:
On May 11, 10:18 am, zdzis1 <zdz... at (no spam) 31.pl> wrote:
On Sun, 11 May 2008 07:33:43 -0700, kT wrote:
On May 11, 7:50 am, zdzis1 <zdz... at (no spam) 31.pl> wrote:
You can ahahaha all you want. Linear is bullshit, quadratic is
bullsht, exponential is bullshit etc. The only model which
wouldn't be bullshit would be an INFORMED model that takes into
account physics. This model has yet to be built.
Cuz you sez so, on the usenet, right?
give me an example of such a model
Here's one, just for starters :
http://www.giss.nasa.gov/research/modeling/gcms.html
Yawn... it is one of the many models produced by this organisation -
none of which works (this is why they keep changing them). If they
did we would they would have been able to predict last summer or this
winter - which they were not. The fact that people talk about physics
while creating their model does not mean that the model is correct.



Why don't you try learning about them before you go shooting off
your mouth and letting everybody on to how clueless you are. Start
here : http://en.wikipedia.org/wiki/Global_climate_model
please ... don't make me lough
I'll try not to, whatever loughing is, it sounds bad.

once again - the fact that people create models does not mean that
they are correct
So, let me get this straight, first you say there are NO models, now
you say the models aren't correct. Why am I not surprised?

Ok, if youz sez souz, on the usenets.

OK, I can agree with you that there are models which include physics.

Ok, that's a start. You admit you were wrong, and it logically follows
that you didn't even bother to look.


I think that they do not work.

But you don't have a clue why, you just think that all you have to do is
SAY that they don't work, and most of the usenet idiots will buy it.

happy?

Don't give up your day job at 'BEST BUY'.
I would first need to get a day job as 'BEST BUY' (I don't even know what

On Mon, 12 May 2008 08:58:50 -0500, kT wrote:
zdzis1 wrote:
On Sun, 11 May 2008 18:09:29 -0700, kT wrote:

On May 11, 5:18 pm, zdzis1 <zdz... at (no spam) 31.pl> wrote:
On Sun, 11 May 2008 15:01:20 -0700, kT wrote:
On May 11, 10:18 am, zdzis1 <zdz... at (no spam) 31.pl> wrote:
On Sun, 11 May 2008 07:33:43 -0700, kT wrote:
On May 11, 7:50 am, zdzis1 <zdz... at (no spam) 31.pl> wrote:
You can ahahaha all you want. Linear is bullshit, quadratic is
bullsht, exponential is bullshit etc. The only model which
wouldn't be bullshit would be an INFORMED model that takes into
account physics. This model has yet to be built.
Cuz you sez so, on the usenet, right?
give me an example of such a model
Here's one, just for starters :
http://www.giss.nasa.gov/research/modeling/gcms.html
Yawn... it is one of the many models produced by this organisation -
none of which works (this is why they keep changing them). If they
did we would they would have been able to predict last summer or this
winter - which they were not. The fact that people talk about physics
while creating their model does not mean that the model is correct.

Why don't you try learning about them before you go shooting off
your mouth and letting everybody on to how clueless you are. Start
here :http://en.wikipedia.org/wiki/Global_climate_model
please ... don't make me lough
I'll try not to, whatever loughing is, it sounds bad.

once again - the fact that people create models does not mean that
they are correct
So, let me get this straight, first you say there are NO models, now
you say the models aren't correct. Why am I not surprised?

Ok, if youz sez souz, on the usenets.

OK, I can agree with you that there are models which include physics.

Ok, that's a start. You admit you were wrong, and it logically follows
that you didn't even bother to look.

I think that they do not work.

But you don't have a clue why, you just think that all you have to do is
SAY that they don't work, and most of the usenet idiots will buy it.

I think it should work the other way - people who claim that their models
work should prove this

so far I have seen model working for past data,

but I haven't seen models predicting future correctly.

There have been a
few papers summarizing past models which show that reality does not want
to agree with the modellers. But, as you say, what do I know?

Perhaps
there are papers showing the prospective value of computer models.

If you
know of any such papers please let me know. Show me papers published,
say, in 2005 which correctly predicted weather in 2007/2008 and I will
publicly admit that such models exist. Really.

happy?

Don't give up your day job at 'BEST BUY'.

I would first need to get a day job as 'BEST BUY' (I don't even know what
it is). My day job is, believe it or not, modeling biological and
biomedical time series. And no, I'm not going to give it up.

I think it should work the other way - people who claim that their
models work should prove this

Von Neumann machines are deterministic, the models run just fine.

What more proof do you need? You don't seem to understand even what a
model is, and how a computer actually works.

so far I have seen model working for past data,

So far you have only demonstrated here that you have ZERO research and
development skills, don't even bother to do the most cursory searches on
the subject before you feel you have something to say about it. You are
an easily identifiable CRACKPOT.
Show me what a cracpot I am and how clever you are - give an example of a

but I haven't seen models predicting future correctly.

It's the future. We don't know what correctly is, indeed, there is no
such thing as correct except in mathematics, and Deterministic von
Neumann machines work just fine in determining correctness of arithmetic
and logic.
All of them within an axiomatic system? And whad does it have to do with

There have been a
few papers summarizing past models which show that reality does not
want to agree with the modellers. But, as you say, what do I know?

You don't know shit by the content of your posts.

Perhaps
there are papers showing the prospective value of computer models.

But you can't be bothered with the trouble of typing keywords into the
search bar, looking at the screen and reading the words, can you.
I have tried, but failed. Even wikipedia says they do not work.

If you
know of any such papers please let me know. Show me papers published,
say, in 2005 which correctly predicted weather in 2007/2008 and I will
publicly admit that such models exist. Really.

So you admit you don't know the difference between weather and climate
as well. Why am I not surprised.
See below.

happy?

Don't give up your day job at 'BEST BUY'.

I would first need to get a day job as 'BEST BUY' (I don't even know
what it is). My day job is, believe it or not, modeling biological and
biomedical time series. And no, I'm not going to give it up.

Somebody has to work the machines. That's you.

Leave the science to people who take the time to read it, and consider
it, and then type letters and words into the search bar. That's not you.

On May 12, 10:30 am, zdzis1 <zdz... at (no spam) 31.pl> wrote:

I think it should work the other way - people who claim that their models
work should prove this

Von Neumann machines are deterministic, the models run just fine.

What more proof do you need? You don't seem to understand even what a
model is, and how a computer actually works.

so far I have seen model working for past data,

So far you have only demonstrated here that you have ZERO research and
development skills, don't even bother to do the most cursory searches
on the subject before you feel you have something to say about it. You
are an easily identifiable CRACKPOT.