Newsgroup sci.stat.math 14318

Directory

In article <5nclsf$kmi$1@news2.digex.net>,
Michael Cohen  wrote:
>Robert Israel (israel@math.ubc.ca) wrote:
>: For example, let A and B be independent normal RV's with mean 0, let 
>: X = A sgn(B) and Y = B sgn(A) (where sgn(x) = 1 if x > 0, -1 otherwise).
>: Then X and Y are both normal, but X+Y is not.
>: In fact it is easy to see that the probability density of X+Y at 0 is 0:
>: since X and Y always have the same sign, the only way for X+Y to be near 
>: 0 is to have both A and B near 0, so Pr(|X+Y|Perhaps this should simply be Y = B.  As originally given, X and Y could
>have different signs, but now they don't.  The argument works.
No, X and B have different signs if A < 0.  X and Y, as I defined them, have 
the same sign: + if A and B have the same sign, - if A and B have different
signs (ignoring values of 0, of course, since that has probability 0).
Robert Israel                            israel@math.ubc.ca
Department of Mathematics             (604) 822-3629
University of British Columbia            fax 822-6074
Vancouver, BC, Canada V6T 1Y4

I need some help with "stats." Using Julian Simon's resampling statistics
program I was asked to estimate the probability of winning the World Series
if the probability of team A winning is .55.  How would I write the program
for this using a histogram and reading the frequency to find out the
probability.
Another problem I was asked to solve using resampling statistics was: 
(parking problems) Estimate the time it takes to get to class and park
using replacement.  Calculate the mean median mode variance and standard
deviation in a sample of 25 and a sample of 100.  
time  frequency
  5     35
10     90
15     70
20     40
25     25
30     30

The question has been answered. The specific application for which
this was of interest is the casino game of blackjack. The "metagames" are
blackjack hands or "rounds", conditioned on a "true count" parameter that
the player calculates at the table, usually as a function of cards seen since
the last shuffle. The restriction of betting from 1 to N units is one that
players frequently impose on themselves in order to avoid unwanted scrutiny
from the pit.
For the solution, see the articles on optimal betting stragegy at
http://www.bj21.com/wgintro.html
           Pete

We have been going back and forth on this for some time, thought
someone here could help...
Okay, take a 52 card deck and turn over three cards.  What are your
odds of turning over a Jack, Seven and/or Two?
We think that the answer is 55%, solving this way:
Because of all the permutations, easier to find when you will not get
one of these cards and subtract from 1.
1 - (40/52)(39/51)(38/50)
Are we right?
Thank you,
Sean Ellis
PS - please post and email

cramer@email.unc.edu (Elliot Cramer) wrote:
>Jim C Chao (e0fdrdde@mail.erin.utoronto.ca) wrote:
>: Hi, I understand it's easy to generate data for a univariate normal
>If y is a vector or n(0,1) variates, Ay has a variance covariance matrix
>AA'.  This can conveniently be a factorization e.g. cholesky of the Var
>matrix you want.
>
this is how I have done it in the past. The Cholesky decomposition is
very easy to implement and is described in virtually any matrix theory
or numerical analysis book. Note that if you want to generate N(mu,
sigma) variables (mu not equal to 0), you need to add mu to the Ay
vector i.e z = Ay + mu has a N(mu, sigma) distribution.
Regards
Joseph