Plot method for stepmented objects

Takes a fitted stepmented object returned by stepmented() and plots (or adds) the fitted piecewise constant lines for the selected stepmented term.

Usage

# S3 method for class 'stepmented'
plot(x, term, add = FALSE, res = TRUE, conf.level=0, interc = TRUE, add.fx = FALSE, 
      psi.lines = TRUE, link=TRUE, const=NULL, res.col=grey(.15, alpha = .4), 
      surf=FALSE, zero.cor=TRUE, heurs=TRUE, shade=FALSE, se.type=c("cdf","abs","none"), 
      k=NULL, .vcov=NULL, leg="topleft", ...)

Arguments

x: a fitted stepmented object.
term: the stepmented variable having the piece-wise constant relationship to be plotted. If there is a single stepmented variable in the fitted model x, term can be omitted.
add: when TRUE the fitted lines are added to the current device.
res: when TRUE the fitted lines are plotted along with corresponding partial residuals.
conf.level: the confidence level for the pointwise confidence intervals for the expected values.
interc: if TRUE the computed components include the model intercept (if it exists).
add.fx: logical. If TRUE and the object fit also includes an additional term for the same stepmented variable, the plot also portrays such `additional' term.
psi.lines: if TRUE vertical lines corresponding to the estimated changepoints are also drawn
link: if FALSE the fitted lines (and possibily the residuals) are reported on the response scale. Ignored if the fit object x is not a glm-like fit.
const: constant to add to each fitted segmented relationship (on the scale of the linear predictor) before plotting. If const=NULL and the fit includes a segmented interaction term (obtained via seg(..,by) in the formula), the group-specific intercept is included.
res.col: when res=TRUE it means the color of the points representing the partial residuals.
surf: if the object fit x includes 2 stepmented covariates (x1 and x2, say) with relevant estimated breakpoints, surf=TRUE will draw on the plane x1-x2 the areas splitted according to the estimated breakpoints with corresponding estimated means superimposed.
zero.cor: see zero.cor in vcov.stepmented; effective only if conf.level>0.
heurs: logical; if TRUE, heuristic (usually somewhat conservative) confidence intervals are computed and plotted; effective only if conf.level>0.
shade: if TRUE the pointwise confidence intervals are portrayed via shaded area; effective only if conf.level>0.
se.type: which standard errors should be computed? see type in vcov.stepmented; effective only if conf.level>0.
k: The value to be passed to vcov.stepmented to computed the standard errors.
.vcov: The estimate var-covariance matrix; if NULL, it is computed internally by vcov.stepmented.
leg: If the plot refers to stepmented relationships in groups, i.e. term has been specified as a vector, a legend is placed at the specified leg position. Put NA not to draw the legend.
...: other graphics parameters to pass to plotting commands: `col', `lwd' and `lty' (that can be vectors and are recycled if necessary, see the example below) for the fitted piecewise constant lines; `ylab', `xlab', `main', `sub', `cex.axis', `cex.lab', `xlim' and `ylim' when a new plot is produced (i.e. when add=FALSE); `pch' and `cex' for the partial residuals (when res=TRUE, res.col is for the color).

Details

Produces (or adds to the current device) the fitted step-function like relationship between the response and the selected term. If the fitted model includes just a single `stepmented' variable, term may be omitted. If surf=TRUE, and res=TRUE the point widths are proportional to the partial residual values.

Value

None.

Author

Vito M. R. Muggeo

Note

Implementation of confidence intervals for the conditional means in stepmented regression is under development; conf.level>0 should be used with care, especially with multiple jumpoints.

Examples


#Following code in stepmented..
if (FALSE) { # \dontrun{

par(mfrow=c(1,3))

plot(os,"x")
plot(os,"z")
plot(os,"z", add.fx=TRUE, psi.lines=FALSE ) 
lines(os, "z")

#display the 'surface'
par(mfrow=c(1,3))
plot(os, surf=TRUE, col=1, res.col=2)
plot(os, surf=TRUE, lty=2)
plot(x,z)
plot(os, surf=TRUE, add=TRUE, col=4, res=FALSE)

} # }