makef: Make non-linear features

In relgam: Reluctant Generalized Additive Models

Description

Internal function for making non-linear features.

Usage

makef(x, r, df, tol = 0.01, removeLin = T)

Arguments

x	Input matrix, of dimension nobs x nvars; each row is an observation vector.
r	Vector of residuals.
df	Degrees of freedom for the fit.
tol	A tolerance for same-ness or uniqueness of the x values. To be passed to the smooth.spline() function. Default is 0.01.
removeLin	If TRUE (default), removes the linear component from the newly created non-linear features.

Value

A list:

f	Non-linear features associated with the features in x.
spline_fit	A list of the spline fits of the residual against each feature. Useful for creating the non-linear features for new data.
lin_comp_fit	If removeLin = TRUE, a list of coefficients for simple linear regression of non-linear feature on original feature. Useful for creating the non-linear features for new data.

relgam documentation built on Jan. 13, 2020, 5:06 p.m.