2D non-polynomial function fitting from the command line -

I have just written a simple Unix command line utility which can be implemented very efficiently. I can measure its performance by running it on several inputs and measuring it while taking it. This will generate a set of pairs of numbers, where S is the size of the input and the processing time. In order to determine the performance characteristics of my utility, I have to fit the function through these data points. I can do it manually, but I like to be lazy and give me a facility.

Is such a facility present?

Its input is the sequence of pairs of numbers. Its output is a formula that indicates that the second number depends on the first one as a function, and additionally an error solution.

One step on the way is for a polynomial which only does for polynomials. This, but it did not produce a ready-made solution.

The next step is to expand the utility to try non-polynomial terms: negative-degree polynomial (as in y = 1 / x) and logarithmic conditions (as in y = x log x ) Should also be tried. To deal with non-polynomial terms, an idea is to surround the polynomial fitting with the X and Y scale transformations. I do not know what he will do related, but not at all.

As I said, I am lazy: I am not looking for ideas on how to write it myself, I am looking for a reliable result of a project which is already my Any suggestions for what?

I believe that in SAS this is RS / 1, I think mathematics This is an old form of Execel and most of the spreadsheets and usually add-ons are available for more advanced forms.

RE, Command Line Tools:

SAS, RS / 1 and Minibab were all command line tools. Years ago when I used them I'm sure at least one of them still has this capability.