Polynomial Features

noun

say it: “pol-ee-NOH-mee-ul FEE-cherz” /ˌpɑliˈnoʊmiəl ˈfitʃərz/

Extra input columns made by multiplying a feature by itself, such as its square, so a straight-line model can follow a curve.

What are polynomial features?

Polynomial features are new columns made by raising a feature to powers, such as rainfall² beside rainfall. A linear model given them can follow a curve while still learning one coefficient per column. scikit-learn's PolynomialFeatures makes them.

from sklearn.preprocessing import PolynomialFeatures
rain = [[2], [3], [4]]
print(PolynomialFeatures(degree=2, include_bias=False).fit_transform(rain))
# [[ 2.  4.]
#  [ 3.  9.]
#  [ 4. 16.]]

In a sentence

  • “With degree=2, polynomial features put each value's square beside it.”
  • “Adding polynomial features let the line bend to follow the harvest curve.”

Level: Journeyman · More AI Basics terms at this level

Updated