modelData: Process the data needed for modelling

In hypergsplines: Bayesian model selection with penalised splines and hyper-g prior

Description

Process the data needed for modelling

Usage

  modelData(y, X, continuous = rep.int(TRUE, ncol(X)),
    nKnots = 4L, splineType = "linear",
    gPrior = c("hyper-g/n", "hyper-g"))

Arguments

y	the numeric response vector
X	the numeric matrix of covariates
continuous	logical vector specifying which covariates really are continous and can be included nonlinearly in a model (default: all covariates are continuous)
nKnots	number of (quantile-based) spline knots (default: 4)
splineType	type of splines to be used (default: “linear”), see makeBasis for possible types.
gPrior	hyperprior for g, either “hyper-g/n” (default) or “hyper-g”.

Value

a list with the internally needed results.

Author(s)

Daniel Sabanes Bove [e-mail]

Examples

## get some data
attach(longley)


## try the function
md <- modelData(y=Employed,
                X=cbind(GNP, Armed.Forces))

## look at the results
str(md)


## try again with cubic splines
md <- modelData(y=Employed,
                X=cbind(GNP, Armed.Forces),
                nKnots=10L,
                splineType="cubic")

## look at the results
str(md)

hypergsplines documentation built on May 2, 2019, 6:14 p.m.