Skip to content

Latest commit

 

History

History
53 lines (33 loc) · 1.45 KB

README.md

File metadata and controls

53 lines (33 loc) · 1.45 KB

Multivariate Polynomial Fit

Holds a python function to perform multivariate polynomial regression in Python using NumPy

See related question on stackoverflow

This is similar to numpy's polyfit function but works on multiple covariates

Origin

This code originated from the following question on StackOverflow

http://stackoverflow.com/questions/10988082/multivariate-polynomial-regression-with-numpy

Author

Matthew Rocklin

License

New BSD license. See LICENSE.txt

Disclaimer

This is not a commonly used method. It often results in a solution with many non-zero coeffieicients like

10 x**2 + 0.01 x y - 0.02 x + 20 y - 0.03 y**2

Instead of a sparse solution like

10 x**2 + 20 y

To obtain sparse solutions (like the second) where near-zero elements are eliminated you should probably look into L1 regularization

Sorry

But I rarely respond to questions about this repository. It is oddly popular but the implementation is pretty dense and so this project generates a large number of reasonable questions. Unfortunately I don't have time to respond to all of these.

I would care more about this project if it contained a useful algorithm. It doesn't.
Read the disclaimer above.