| Date: | Tue, 2 Feb 1999 20:48:57 -0800 |
|---|
| Reply-To: | rhpcvn@tiac.net |
|---|
| Sender: | "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU> |
|---|
| From: | Richard Palmer <rhpcvn@TIAC.NET> |
|---|
| Organization: | MAPS |
|---|
| Subject: | Re: Simulating A Distribution |
|---|
| Content-Type: | text/plain; charset=us-ascii |
|---|
Simulating an Unknown Continuous Distribution doesn't present any new
problems if there are enough observations. Make a histogram of the
continuous observations and use the frequencies to do a simulation of a
discrete distribution. You can add some "frills" like returning a
uniform (not normal) random number over the range of the histogram bar
rather than the midpoint.
Richard Palmer
|
