Slimy, Rich continues his OT anti-Apple rants.

On 9/2/2011 3:21 PM, Wolfgang Weisselberg wrote:
wrote:
On 8/17/2011 11:26 AM, Wolfgang Weisselberg wrote:
wrote:
On 8/14/2011 6:43 AM, Wolfgang Weisselberg wrote:
tony wrote:

Oh, I think I can say that with about 99% probability of accuracy.

You are saying lots of wrong things lately. For example, nospam's
pretty convincing to me. "anyone" is thus wrong by default, so
the accuracy is ZERO. So your 99% happens to not capure reality.

Did you understand that market position is considered on a periodic basis.

I'll gift you with an '?'.

If market position is on a periodic basis how is period of sale
accounted for by casual observation?

If tony can use population numbers (see other posts) ---
and ones sampled with a easily strong bias --- to give market
share numbers, then so can I. A higher number of Mac laptops
in population compared to their sales per time-unit rate
would show that they are used longer, hence cheaper to buy as
price-per-time-unit-of-ownership than the price at the time of
purchase would indicate in a naive comparison.

Anyway, we're arguing over a factor 5, and I'd guess even Mac
laptops are laid to rest after twice the age of a Windows laptop.
Still a 2.5x difference, and a good idea to buy a Mac laptop even
if it would be quite a bit more expensive on identical features
than Windows laptops.

You didn't answer

You didn't bother to read the answer. You didn't even bother
to end your sentence with a '.'. You're just trying to be a
nuisance.

-Wolfgang

You still haven't answered

It's such a pleasure to engage in a conversation with you As you use
punctuation as an excuse to avoid answers


Here are enough periods for the next month
.................................................. .................................................. ................



--
Peter

On 9/2/2011 3:24 PM, Wolfgang Weisselberg wrote:
wrote:
On 8/17/2011 2:26 PM, Wolfgang Weisselberg wrote:
wrote:
On 8/16/2011 1:11 PM, Wolfgang Weisselberg wrote:
wrote:
On 8/13/2011 5:58 PM, Wolfgang Weisselberg wrote:
wrote:
On 8/9/2011 6:09 AM, Wolfgang Weisselberg wrote:
wrote:


Product design is a part of every relevant part of a company,
usually including decision makers way above marketing. Marketing
gives input, but so do others, because often marketing doesn't
grasp the lack of unobtainium.

Meaning?

*sigh* Obviously you never had the very common experience of
sales wanting things that are physically impossible. Or
simply not possible within the timeframe given.

I see you silent.
Maybe you googled?


You claimed predictive analysis needed accurate numbers. You have
neither proven that they are accurate (Apple's sales numbers aren't
exactly diapers and beer) nor that they need to be very exact.

correct accurate to margin of error, not inaccurate.

I see. So, say, 5% ±95% would be would be accurate to the
margin of error and very worthless.

I see you silent.
Perhaps you understood that your "accurate number" wasn't very
exact, argument wise.


No predictive analysis can be more accurate than the numbers upon which
it is based.

I see. That's why averaging samples could never work, right?

I see you silent.
And I had so hoped to tell you binning works, for example.
But you found out yourself that you wrote an indefensible claim.


that's why it is also referred to a probability analysis.

Really! Because it cannot be more accurate than ... and not
because it tries to estimate probabilities or sumtin' like
that.

I see you silent.
Hmmm. Could it be I hit the nail on the head?


If I accurately observe that a fair coin has landed on tails, 200,000
time in a row, the probability of it landing on tails is exactly .5.
Prior events have nothing to do with the probability of the next event.

And that's why a probability analysis needs accurate data of
past events!

Again, I see you silent. Probably because you you found your
straw argument is silly and indefensible.


Actually, there are a lot of things where the probability of
the next event is determined by past events. For one trivial
example, the probability of pregnancy also depends on the past
random event of you being born male or female.

Going from a fair coin to gender is a straw argument. Gender analysis
fair coin analysis.

It's so very typical of you to only look on the very surface.
(Probably because it's the only way out for you now.)


For example, taking umbrage at the very surface of a counter
example ... which uses gender, as it's quite intuitive to most
people that males don't often get pregnant.


I could as well have used Russian roulette (the chance of being
killed depending on the random events of the shot being fired
or not fired n times before you're on the spot). Or drawing
items out of a receptable without putting them back. Or drawing
multiple cards (without putting them back) from a stack of cards.

Here's something for you to read on that stuff:
http://www.intmath.com/counting-prob...robability.php
http://www.intmath.com/counting-prob...ent-events.php

"Prior events have nothing to do with the probability of the
next event" the way you stated it as universally valid is UTTER
BULL****, because that's only true for independent events.
Many events, probably most events, are anything but independent.

-Wolfgang

You are deliberately ignoring my comments, or evading what I said.



--
Peter

On 9/2/2011 3:20 PM, Wolfgang Weisselberg wrote:
wrote:
On 8/17/2011 11:23 AM, Wolfgang Weisselberg wrote:
wrote:
On 8/13/2011 5:57 PM, Wolfgang Weisselberg wrote:

Above laptop inbuild in the car is counted as Windows laptop
in market share numbers?

Please define what you mean by laptop. Can you take this laptop out of
the car and take it on a plane.

You can even take the car on a plane. You can also just take
the right front wheel of the car into a plane. So what gives?

I see you have no answer. Thought so.


If so, what process is involved in the
removal?

Is that relevant? A car radio can be removed in mere seconds ...
a wheel too, given the right tools.

Given the right tools would you then be capable of rational and
practical thinking????????????????

Either you are a bad copy of Megahal or you quaffed too much
of Whiskey-Dave's stuff. Anyway, you don't make any sense,
and you know it.


Hope I didn't use up my quota of question marks for the day.

You need a new keyboard, your '?' key has terrible contact
chatter.

-Wolfgang

I have no interest in continuing your bull****.
bye


--
Peter

On Fri, 2 Sep 2011 21:23:44 +0200, Wolfgang Weisselberg
wrote:

tony cooper wrote:
On Wed, 17 Aug 2011 20:04:41 +0200, Wolfgang Weisselberg
wrote:

Any statistical sampling analysis is by definition not exact as it
contains a desired degree of uncertainty.

Actually, statistical sampling projections have a desired degree of
certainty, not uncertainty. The reciprocal between the degree of
certainty and total accuracy is the uncertainty factor.

Wrong. There is at least one case where the sampling is
exact: namely when you sample every case.

You'd think that even Wolfie would understand what "sample" means
statistically. "Sample every case" is an oxymoron.

So prove me wrong by quoting a relevant statistics text.

If you can't figure out that a sample is a subset of a population and
the entire population is a different thing, then I can't help you.


You would also think that even Wolfie would understand that not all
responders provide accurate and true responses, so even a 100% survey
does not result in complete certainty.

Where in the world does tony get the abstruse ideas then that
- all laptop manufacturers and sellers provide accurate and
true data

You're mixing things up. The discussions about laptop market share
figures was not based on sampling techniques or surveys. Those
figures are based on submissions of figures by the manufacturers.

There's no guarantee of accuracy, but it is to each submitter's best
interest to submit accurate figures. They are submitting the figures
in order to know what other manufacturers are doing. If they abuse
the system, the other manufacturers will abuse the system and no one
gains.

- there are only cases that involve asking responders, and no
cases where e.g. counting people or cars etc. can be used
for statistics?

Yes, and examples of that were given in an earlier post. I used the
example of an automobile dealership having a survey conducted that
counted cars on a particular road to determine the benefit of buying
billboard space on that road.

Foot traffic surveys, and road traffic surveys, are frequently used by
firms involved in renting retail space or selling commercial property.
If you have a commercial building for sale or rent, it makes sense to
say "X thousand people pass this space every day".

What with extrapolation election results on partially (e.g. counted
in some districts, but not yet in others) counted data? Who's the
responder there?

You talking about Exit Polls? People leaving the poll site are asked
who they voted for.


--
Tony Cooper - Orlando, Florida

On 9/2/2011 5:28 PM, tony cooper wrote:
On Fri, 2 Sep 2011 21:23:44 +0200, Wolfgang Weisselberg
wrote:

tony wrote:
On Wed, 17 Aug 2011 20:04:41 +0200, Wolfgang Weisselberg
wrote:

Any statistical sampling analysis is by definition not exact as it
contains a desired degree of uncertainty.

Actually, statistical sampling projections have a desired degree of
certainty, not uncertainty. The reciprocal between the degree of
certainty and total accuracy is the uncertainty factor.

Wrong. There is at least one case where the sampling is
exact: namely when you sample every case.

You'd think that even Wolfie would understand what "sample" means
statistically. "Sample every case" is an oxymoron.

So prove me wrong by quoting a relevant statistics text.

If you can't figure out that a sample is a subset of a population and
the entire population is a different thing, then I can't help you.


You would also think that even Wolfie would understand that not all
responders provide accurate and true responses, so even a 100% survey
does not result in complete certainty.

Where in the world does tony get the abstruse ideas then that
- all laptop manufacturers and sellers provide accurate and
true data

You're mixing things up. The discussions about laptop market share
figures was not based on sampling techniques or surveys. Those
figures are based on submissions of figures by the manufacturers.

There's no guarantee of accuracy, but it is to each submitter's best
interest to submit accurate figures. They are submitting the figures
in order to know what other manufacturers are doing. If they abuse
the system, the other manufacturers will abuse the system and no one
gains.


Because Apple is a publicly traded security, the knowing release of
inaccurate information could be a violation of Securities laws.


--
Peter

On Fri, 2 Sep 2011 21:22:43 +0200, Wolfgang Weisselberg
wrote:

Handwaving. Lots of. And you're wrong, to boot: there's an
equally generally accepted definition for market share that is
in monetary units.

Where in the world do you get that idea?

Market share can be expressed in dollars or in units depending on what
is to be determined.

A fast food restaurant is interested in another fast food's market
share in dollars because their product line is comparably priced and
they are going after the same market.

An auto maker is interested in another auto maker's unit sales because
the product line may be priced differently, the price includes
after-market items, and the price is affected by the dealer's
policies.

Also you don't seem to understand that we're not trying to
measure market share, we're falsifying a given number as
market share. All the reasons why Mac laptops could be seen
more often don't explain the discrepancy between what's
expected and what's observed.

By one observer with a limited exposure in a limited environment and a
limited view of who owns what?


--
Tony Cooper - Orlando, Florida

On 9/2/2011 5:28 PM, tony cooper wrote:
On Fri, 2 Sep 2011 21:23:44 +0200, Wolfgang Weisselberg
wrote:

tony wrote:
On Wed, 17 Aug 2011 20:04:41 +0200, Wolfgang Weisselberg
wrote:

Any statistical sampling analysis is by definition not exact as it
contains a desired degree of uncertainty.

Actually, statistical sampling projections have a desired degree of
certainty, not uncertainty. The reciprocal between the degree of
certainty and total accuracy is the uncertainty factor.

Wrong. There is at least one case where the sampling is
exact: namely when you sample every case.

You'd think that even Wolfie would understand what "sample" means
statistically. "Sample every case" is an oxymoron.

So prove me wrong by quoting a relevant statistics text.

If you can't figure out that a sample is a subset of a population and
the entire population is a different thing, then I can't help you.


Shame it's too much effort to do a Venn diagram in a binary group.
Anyway as far as I am concerned there is not point in continuing with him.




--
Peter

On 9/2/2011 5:45 PM, tony cooper wrote:
On Fri, 2 Sep 2011 21:22:43 +0200, Wolfgang Weisselberg
wrote:

Handwaving. Lots of. And you're wrong, to boot: there's an
equally generally accepted definition for market share that is
in monetary units.

Where in the world do you get that idea?

Market share can be expressed in dollars or in units depending on what
is to be determined.

A fast food restaurant is interested in another fast food's market
share in dollars because their product line is comparably priced and
they are going after the same market.

An auto maker is interested in another auto maker's unit sales because
the product line may be priced differently, the price includes
after-market items, and the price is affected by the dealer's
policies.

Also you don't seem to understand that we're not trying to
measure market share, we're falsifying a given number as
market share. All the reasons why Mac laptops could be seen
more often don't explain the discrepancy between what's
expected and what's observed.

By one observer with a limited exposure in a limited environment and a
limited view of who owns what?



I strongly suspect he's trolling you.

--
Peter

PeterN wrote:

You totally miss that in today's economy with JIT availability,
production is a function of actual sales, which is a determinant of
market share.

Please provide proof for that claim. Let's start --- this being
a photo newsgroup --- with lenses. Which, as far as I know, are
produced in large batches well ahead of the eventual sales and well
ahead of knowledge of sales numbers, especially for the first run.

For example, I can buy the Canon 70-200mm f/2.8 IS version I ...
http://www.amazon.com/Canon-70-200mm.../dp/B00006I53X
.... new. Even though it's been replaced by version II --- in
March 2010, 17 months ago. Now, is *that* JIT to you?



Please feel free to explain (in detail!) how lead times of at the
very least 17 months (and more likely 2-4 or even more years,
Canon has no extra assembly line for every single lens they
sell!) are JIT; or just admit you were wrong.

-Wolfgang

PeterN wrote:

You totally miss that in today's economy with JIT availability,
production is a function of actual sales, which is a determinant of
market share.

That would explain why it takes *months* for new e.g. Canon cameras
to appear here, that would explain why e.g. Canon produces every
single lens just in time.

Did you know that Canon is spending more time retooling the
production lines than producing lenses and cameras? I told them
to produce a stockpile that will hopefully(!) sell over time
(sometimes even over years) and actually spend most of the time
*producing*, but they just muttered "PeterN said we have to
do JIT".


Not stockpiling only works when your demand is identical to your
production line capacity or higher.

-Wolfgang