BasisBuilder

class liesel_gam.BasisBuilder(registry, names=None)

Bases: object

Initializes Basis objects from data in a PandasRegistry.

Parameters:

• registry (PandasRegistry) – A pandas registry, giving access to the data.

• names (NameManager | None, default: None) – A name manager for creating unique names.

See also

TermBuilder

Initializes structured additive terms.

Basis

Basic basis class.

LinBasis

Specialized basis for linear effects.

MRFBasis

Specialized basis for Gaussian Markov random fields.

Examples

>>> import liesel_gam as gam
>>> df = gam.demo_data(n=100)
>>> registry = gam.PandasRegistry(df)
>>> bb = gam.BasisBuilder(registry)
>>> bb.ps("x_nonlin", k=20)
Basis(name="B(x_nonlin)")

Methods

basis	Initializes a general basis given a basis function.
bs	B-spline basis with integrated squared derivative penalties.
cc	Cyclic cubic regression spline basis and penalty matrix.
cp	Cyclic P-spline basis and penalty matrix.
cr	Cubic regression spline basis and penalty matrix.
cs	Cubic regression spline basis and penalty matrix with null space penalty.
kriging	Gaussian process models with a fixed range parameter in a basis-penalty-parameterization, often referred to as Kriging.
lin	Linear design matrix without penalty.
mrf	Gaussian Markov random field basis and penalty.
ps	B-spline basis with a discrete (P-spline) penalty matrix.
ri	Random intercept basis.
tp	Thin plate spline basis and penalty matrix.
ts	Thin plate spline basis and penalty matrix with null space penalty.

Attributes

data	The dataframe wrapped by this builder's registry.