Search code examples
pythonpython-3.xnumpypolynomials

NumPy Polynomial Fitting

I have the following lines of code

import numpy as np
from numpy.polynomial import Polynomial

x=Polynomial.fit([0,1,2,3,4],[0,1,2,3,4],4)
print(x)
print(x.call(0))

I would expect numpy to fit the data to the polynomial f(x)=x. But instead it outputs 2.0 + 2.0 x + (1.2915422e-15) x2 - (7.05812123e-16) x3 - (9.57512621e-17) x**4.

Even stranger when i get it to evaluate at x=0 it gives 1.1102230246251565e-15 which is what i would expected with f(x)=x but isn't 2.0. Does anyone know what is going on?

Using the old polynomial features of numpy i.e. np.polyfit I get a correct answer but the rest of my project uses the new features.

Solution

  • If you read the docs, you see that what fit returns is not literally the coefficients. If you print(x.convert().coef), you'll see that it is, indeed, f(x) = x:

    [-1.77635684e-15  1.00000000e+00  1.22916662e-15 -3.08599735e-16
  1.95879353e-17]

    https://numpy.org/doc/stable/reference/generated/numpy.polynomial.polynomial.Polynomial.fit.html