Preprocessing 0.2.0 polynomialfeatures

Lodestar.Preprocessing 0.2.0. This page is frozen at that release. Read the current documentation for what main says now. A link to a decision or a migration page follows main, and leaves the archive.

Home › Feature transforming

PolynomialFeatures

Expands each row into its polynomial terms, at sklearn.preprocessing.PolynomialFeatures parity.

Fits nothing — the terms are decided by the feature count and the degree, not by the data — so this is static, as Normalizer is.

The column order is the contract. A caller that fits a model on these columns and reads its coefficients back needs to know which coefficient belongs to which term, and the only useful answer is the reference's own order: the bias, then every term of degree one in feature order, then every term of degree two as x0², x0·x1, x1², and so on. FeatureNames returns exactly that order, in the reference's own spelling.

Members

Member	What it does
PolynomialFeatures.Transform	Expands a matrix into its polynomial terms.
PolynomialFeatures.FeatureNames	Each term's name, in the order the columns come out.
PolynomialFeatures.OutputFeatureCount	How many terms an expansion produces, without producing one.