On 18 Jun 2002 07:55:59 -0700, [EMAIL PROTECTED] (Anthony Yip)
wrote:

> hi,
> 
> thanks for the response. unfortunately, the exact steps are still
> unclear to me.. sorry for the trouble. but let's say using the
> standard best fit technique i get the slope of 0.5 and intercept of 40
> for a set of data points. how i would like to force the intercept to
> be 40.04 and have it best fit the data with this constraint.. what are
> the steps i should take? if you can make it unncessarily clear to me
> that would be great. thanks in advance.

I think you should accept the advice of experts and 
avoid trying to perform a statistical fit "with this constraint."

The experts will tell you that fitting with that sort of constraint
is something that *hardly anybody*, out in the average world,
has justification to do.   And the experts will tell you that folks
who don't understand how to do it, backwards and forwards,
are *almost always*  going to belong to the majority who should
not try.  

Further, if the person doesn't understand it thoroughly, 
the fact of 'ignorance' gives a big boost to the idea that 
they ought to keep away --
 a)  Even if their solution *properly* is improved a bit
by forcing the intercept, *usually*,  very little is gained.  
 b)  If it is *improper*  to force the intercept, then the solution
is usually *horrible*, and the tests are totally wrong, too.

Hope this helps.
-- 
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to