Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: [Phys-l] heat loads

Great!

Nope, not for Apple:

* Windows 95/98/ME/2000/NT/XP (There is no version of NLREG for
Linux, Macintosh, Windows 3.11, OS/2)
* At least 64 MB of memory (256 MB recommended).
* 5 MB of disk space.


It's not high cost.

Depending on features. $140 => $245



http://www.nlreg.com/

bc

p.s. This is, except for the distrib. of residuals, Peter's. In his, tho, one must write a C module including derivatives wrt the parameters, as does one w/ kaleidograph (optional, but quicker).


Brian Whatcott wrote:

At 06:16 PM 2/18/2006, you wrote:

another "law": Newton's law of cooling.

2) There is a body of opinion that prefers not to massage data
in order to run a straight line through it.
The least squares fit exaggerates the weight of the cool end.
Better in my view to run a non linear lms fit.


This seems much work. I prefer to massage by using my calibrated
outlier removal judgment.
The first reference I found suggested least trimmed squares might be better?

bc, eliminates outliers one by one until happy.

p.s. got any suggested programs for power mac?



We all have our sweet heart procedures. Mine is NLREG.
It's not high cost. Versions for PC of course, and I think, mac too.
You write the target equation. You provide the list of x & y values.
The program sizes the parameters provided, and shows the expected
normal distribution of the errors. Outliers are easy to pick up.
It supports its results with statistics.



Brian Whatcott Altus OK Eureka!

_______________________________________________
Phys-l mailing list
Phys-l@carnot.physics.buffalo.edu
https://carnot.physics.buffalo.edu/mailman/listinfo/phys-l