fiducial.spline: Sample from fiducial b-spline distribution

In dereksonderegger/fiducialSplines: Fiducial inference for free-knot B-splines in 1 Dimension

Description

Sample from fiducial b-spline distribution

Usage

fiducial.spline(x, y, degree, num.knots, sigma.jitter = NULL,
  sigma.split = NULL, chain.length = 500, start.point = NULL,
  burn.in = 100, num.chains = 3, num.jacobians = 100)

Arguments

x	Vector of x-values
y	Vector of y-values
degree	Degree of the b-spline to fit to the data
num.knots	A scalar or vector of the number of knot points to be considered. e.g. 3:5 or just 3.
sigma.jitter	MCMC tuning parameter: the size of jumps for the knot point locations.
sigma.split	MCMC tuning parameter: how far apart knots end up when we split one.
chain.length	How long a MCMC to create.
start.point	A list of starting points
burn.in	The length of chain to be considered 'burn in'.
num.chains	How many chains to create.
num.jacobians	How many randomly selected jacobians to average for each step of the MCMC.

Examples

require(SemiPar)
data(lidar)
x <- lidar$range
y <- lidar$logratio
plot(x, y)
foo <- fiducial.spline(x, y, 3, 2)
plot(foo)  # plot the three chains...
confint(foo)

dereksonderegger/fiducialSplines documentation built on May 15, 2019, 5:04 a.m.