bentcableAR (version 0.3.0)

stick.ar.0: Broken-Stick Regression for Independent Data

Description

This function is the main engine for bentcable.ar when a broken stick (i.e. $\gamma$=0 for bent cable) model is assumed for independent data. For AR(p) time-series data, this function is intended for determining an appropriate p and initial values for the stick parameters.

Usage

stick.ar.0(init.vect, y.vect, t.vect = NULL, n = NA)

Arguments

init.vect

A numeric vector of initial values, in the form of c(b0,b1,b2,tau).

y.vect

A numeric vector of response data.

t.vect

A numeric vector of design points, which need not be equidistant. Specifying t.vect=NULL is equivalent to specifying the default time points c(0,1,2,...).

n

Length of response vector (optional).

Value

fit

An nls object that is the maximum likelihood fit.

y, t, n

As supplied by the user.

p, stick

The values 0 and TRUE, respectively; used internally by bentcable.ar and cable.ar.0.fit.

Details

The returned object is compatible with a cable.ar.p.iter object for independent data.

The broken stick as a special case of the bent cable has form $f(t) = b_0 + b_1 t + b_2 (t-\tau) I\{t>\tau\} $.

Broken-stick regression by maximum likelihood for independent data is performed via nonlinear least-squares estimation of $\theta=(b_0,b_1,b_2,\tau)$ through the built-in R function nls. The estimation relies on the user-supplied initial values in init.

References

See the bentcableAR package references.

See Also

cable.ar.p.iter, fullcable.t, cable.fit.known.change.

Examples

data(sockeye)

stick.ar.0( c(13,.1,-.7,12), sockeye$logReturns )