​

​PolySurf  generates polynomial models from data. 
It generates a bunch of polynomial combinations, makes data fit 
and selects the best 10 polynomial models. 

The criterion is the sum of residual squares divided by the degrees of freedom.

PolySurf can give an accurate approximation, when the investigated phenomenon is well behaving. 

 

Then the same data can be run with FuncFit to get a function combination formulas for the phenomenon.
FuncFit regards  the polyn0 as the Taylor polynomial of the best function.

​

There may be measurements that do fit too badly in the pattern, normally called
as outliers. So they can be screened away. The decision is left to the user,
who can verify if the data are correct.

 

The text file below contains an example of PolySurf process in cloud service.