Re: 4 parameter beta equivalence
- From: "Samik R." <samik@xxxxxxxxxxxxx>
- Date: Fri, 06 Oct 2006 11:33:48 -0600
On 10/6/2006 10:03 AM, Lou Thraki wrote:
Samik R. wrote:Hello,
I have a seemingly simple question - though I wasn't able to solve
myself. Using Beta(min, max, alpha, beta) as the notation for the
4-parameter beta distribution, the following two same?
1. Beta(min, max, alpha, beta)
2. Beta(0,1,alpha,beta) * (max-min) + min
Most will probably realize that the second one is the formula for PERT
distribution used in project management.
My algebraic manipulations seemed to lead to the following:
PDF of (1):
(x - min)^(alpha-1) * (max - x)^(beta-1)
----------------------------------------
B(alpha,beta) * (max - min)^(alpha+beta-2)
PDF of (2):
x^(alpha-1) * (1-x)^(beta-1)
----------------------------- * (max - min) + min
B(alpha,beta)
If you generate x in [0,1] following Beta(0,1,alpha,beta),
and you perform the transformation y = (max-min)*x + min,
then the variable y will be distributed in [min,max]
following PDF (1).
Does that mean they are equivalent?
Rephrasing, if I generate a set of 100 uniform random numbers between (0,1), and pass them through the inverse CDF of (1) and (2), will I get the same values?
.
- Follow-Ups:
- Re: 4 parameter beta equivalence
- From: Lou Thraki
- Re: 4 parameter beta equivalence
- References:
- 4 parameter beta equivalence
- From: Samik R.
- Re: 4 parameter beta equivalence
- From: Lou Thraki
- 4 parameter beta equivalence
- Prev by Date: Re: Interesting (?) Property of Least Square Line Fit to Evenly-Spaced Data.
- Next by Date: Re: 4 parameter beta equivalence
- Previous by thread: Re: 4 parameter beta equivalence
- Next by thread: Re: 4 parameter beta equivalence
- Index(es):
Relevant Pages
|
