[SciPy-User] multidimensional polynomial fit

Oscar Gerardo Lazo Arjona algebraicamente@gmail....
Sun Jun 13 00:28:28 CDT 2010


 <josef.pktd <at> gmail.com> writes:

> Assuming I understand correctly,  fitting the last variable to a
> polynomial of the first three
> 
> depends on how many cross terms you want.

Well, I already wrote a generalized function that works like polyfit:

def polynomial_fit(points,degree,depreciation=False):
    ...
    ...

It returns an array of the coefficients of the multidimensional polynomial with
the degrees indicated as a list of integers (just as polyfit).

It was quite a lot of work, it's probably the most abstract thing I've done. I
had lot's of fun writing it, and I would like it to be included in numpy (if you
think that is wise).

> here is an example which restricts the powers in the cross-terms
> 
> >>> x = np.arange(5)[:,None]+ [0,10,100]
...
...
> use of OLS can be replaced by np.linalg.lstsq

Well, thank you, but that's a lot more complicated to remember ;)

thanks.

Oscar



More information about the SciPy-User mailing list