cn: Control Nets

Description

Generate the control net for a univariate B-spline

Usage

cn(x, ...)
# S3 method for cpr_bt
cn(x, theta, ...)
# S3 method for formula
cn(
  formula,
  data,
  method = stats::lm,
  method.args = list(),
  keep_fit = TRUE,
  check_rank = TRUE,
  ...
)

Value

a cpr_cn object. This is a list with the following elements. Some of the elements are omitted when using the cn.cpr_bt method.

cn: the control net, data.frame with each row defining a vertex of the control net
bspline_list: A list of the marginal B-splines
call: the call
keep_fit: logical, indicates if the regression model was retained
fit: if isTRUE(keep_fit) then the regression model is here, else NA.
coefficients: regression coefficients, only the fixed effects if a mixed effects model was used.
vcov: The variance-covariance matrix for the coefficients
loglik: The log-likelihood for the regression model
rse: the residual standard error for the regression model

Arguments

x: a cpr_bt object
...: pass through
theta: a vector of (regression) coefficients, the ordinates of the control net.
formula: a formula that is appropriate for regression method being used.
data: a required data.frame
method: the regression method such as lm, glm, lmer, etc.
method.args: a list of additional arguments to pass to the regression method.
keep_fit: (logical, defaults to FALSE). If TRUE the regression model fit is retained and returned in the fit element. If FALSE the regression model is not saved and the fit element will be NA.
check_rank: (logical, defaults to TRUE) if TRUE check that the design matrix is full rank.

Details

cn generates the control net for the given B-spline function. There are several methods for building a control net.

Examples

Run this code


acn <- cn(log10(pdg) ~
              btensor(   x     = list(day, age)
                      , df     = list(30, 4)
                      , bknots = list(c(-1, 1), c(44, 53))
              )
           , data = spdg)
str(acn, max.level = 1)