LS fit with polynomial ratio

[simplemaan]

My application needs to fit data in a single variable according to a ratio of polynomials (a rational fraction).  I would like to be able to specify the polynomial functions and solve for coefficients, not just supply X & Y data points.

The closest solution I can see is the General LS Linear Fit where I can supply the independent variable, x, and the H matrix, but it appears that I can use it to fit either the numerator polynomial or the denominator, but not both at the same time.

Thanks in advance for any help.

[altenbach]

Once you have a ratio of polynomials, the "linear" part in "general LS linear Fit" no longer applies.

You probably could use Levenberg Marquardt (Nonlinear curve fit). Define the model function as the ratio of two polynomials. You can define the polynomial order of each polynomial via the "data" variant input for a quite generic solution to the problem. Of course you would need to supply reasonable parameter estimates.

Do you have some example data and typical orders for the polynomials? How many fitting paramters?