Re: Simulating normally distributed data with constraints
- From: Ray Koopman <koopman@xxxxxx>
- Date: Thu, 16 Aug 2007 16:17:11 -0700
[I'm reading this in Google Groups, which shows that m00es replied
12 hours ago but doesn't give what he said, so please excuse any
redundancies.]
On Aug 16, 3:06 am, perreaul...@xxxxxxxxx wrote:
Hi Folks:
I would like to know a practical way to simulate data which are
normally distributed (bell shape curve), but with the following
constraints:
1- I need discrete interval data (I want an histogram, with categories
on the x axis going from say 1 to 500 (thus 500 bars on my histogram)
2- I want a particular number of observation (say 2000....the total
lenght of all the bars, as measure with the y-axis units, sould be
2000)
I just need to simulate some data set with these requirements, so I'm
looking for a quick and easy way to do it.
Thanks a lot,
Charles P.
If you can generate Uniform[0,1] variables, then try
X = Floor[251 + 250*(U^.135 - (1-U)^.135)],
where U is Uniform[0,1], and Floor[.] means to round the argument
down to an integer, dropping any fractional part. X will be an
integer in [1,500] whose distribution is approximately normal with
mean approximately 250.5 and standard deviation approximately 49.4.
.
- Follow-Ups:
- References:
- Simulating normally distributed data with constraints
- From: perreault . c
- Simulating normally distributed data with constraints
- Prev by Date: Re: Sports probability question
- Next by Date: Re: Moment-Generating-Function for Multivariate Probability Distributions
- Previous by thread: Re: Simulating normally distributed data with constraints
- Next by thread: Re: Simulating normally distributed data with constraints
- Index(es):
Relevant Pages
|
