After considering this problem, this is exactly what I ended up doing. Thank
you very much for the response. Allyn
-----Original Message-----
From: A UK-based worldwide e-mail broadcast system mailing list
[mailto:[log in to unmask]]On Behalf Of Institute for Statistical
Engineering
Sent: Sunday, June 29, 2003 5:10 AM
To: [log in to unmask]
Subject: FW: Evaluating fit of curve to histogram
-----Original Message-----
From: A UK-based worldwide e-mail broadcast system mailing list
[mailto:[log in to unmask]]On Behalf Of Jay Warner
Sent: Saturday, June 28, 2003 6:13 PM
To: [log in to unmask]
Subject: Re: Evaluating fit of curve to histogram
1) You cannot tell if they are 'identical." You can tell if they are
not similar, at some level of confidence. I.e., If you were to collect
your data all over again, in the same way, the probability that you would
get a fit (mse) this small or smaller would be p. Or p% if you like.
2) One way to do this is to break up the theoretical fit into bins of
the same widths, compute the area under the curve for each theoretical bin
(i. e., p of getting a value between these bin limits), then compare these
theoretical p values for each bin against the observed histogram values,
using a chi square test. Not the most sensitive test, but a conservative,
well understood one.
Cheers,
Jay
Allyn Baskerville wrote:
> I have a histogram composed of 24 bins that I've fit a curve to based on
> mean squared error. How do I determine if the curve and histogram are
> identical? Thank you. Allyn
--
Jay Warner
Principal Scientist
Warner Consulting, Inc.
4444 North Green Bay Road
Racine, WI 53404-1216
USA
Ph: (262) 634-9100
FAX: (262) 681-1133
email: [log in to unmask]
web: http://www.a2q.com
The A2Q Method (tm) -- What do you want to improve today?
|