Form of polynomials produced by regress

[Raf.D]

Hello everyone, I trying to fit a function to some data I have. I have 3 independent variable so the equation will be of the form f(x,y,z). Using the Regress interp functions I have created the wanted equation and the accuracy is good enough. My problem is that I need to present this equation. I know that the coefficients are in the outcome of the regress function (after the first three) but given that my polynomial is of 8th or 9th degree there are a couple of hundred coefficients ( but not a perfect cube which would make sense to me). So I will need to present my equation as a sum, but I don't know how the form of the created polynomial to write it as a sum. Anyone know anything about the forms of polynomials created by regress?
Thank you all for your help.

 

[PNachtwey]

You are assuming that a general polynomial will approximate the function as a result of one variable.
I have found that there is a practical order to the degree of polynomial that makes sense.
Start with 3 general polynomials
x0+x1*x+x2*x^2+x3*x^3...
Y0+y1*y+y2*y^2+y3*y^3...
z0+z1*z+z2*z^2+z3*z^3...
and multiply them together. This is easy in Mathcad.

We don't know the range of each variable. Knowing that would help or even getting a clue as to what you are really doing would help.
It could be a different type of polynomial is required for each direction. One could try Pade approximants too.




Peter Nachtwey
Delta Computer Systems

http://www.deltamotion.com

http://forum.deltamotion.com/

IFPE Hall of Fame Member