Primarily conceived for exposition purposes, for the two-parameters case.
The black-filled points are those for which the observed summary statistic was outside of the convex hull of the simulated empirical distribution. The crosses mark the estimated ML point and the confidence intervals points, that is, the outmost points on the contour defined by the profile likelihood threshold for the profile confidence intervals. There is a pair of CI points for each interval.
The smaller black dots mark points added in the latest iteration, if refine was used.
# S3 method for SLik_j
plot(x, y, filled = nrow(x$logLs)>5000L, decorations = NULL,
color.palette = NULL, plot.axes = NULL,
plot.title = NULL, from_refine=FALSE, plot.slices=TRUE,
show_latest=FALSE, ...)
# For objects returned by the primitive workflow:
# S3 method for SLik
plot(x, y, filled = FALSE, decorations = NULL,
color.palette = NULL, plot.axes = NULL,
plot.title = NULL, plot.slices=TRUE, ...)
plot.SLik_j returns invisibly a list including coordinates of the plot(s) (at least if the latest version of the spaMM package is installed).
The exact format will depend on the nature of the plot but the names of elements should be self-explanatory.
plot.SLik returns the plotted object invisibly.
An object of class SLik or SLikp
Not used, but included for consistency with the plot generic.
whether to plot a mapMM or a filled.mapMM.
Graphic directives added to the default decorations value in calls of mapMM or a filled.mapMM (see the source code of plot.SLik for the latter default values).
Either NULL or a function that can replace the default color function used by plot.SLik. The function must have a single required argument, giving the number of color levels. See Details for comments about default value.
statements which replace the default titles to the main plot (see Details).
statements which replace the default axes on the main plot (see Details).
For programming purposes, not documented.
boolean: whether to plot “slices” of the summary-likelihood surface for pairs of parameters (p1,p2), when more than two parameters are fitted. In such plots the additional parameters p3, p4... are fixed to their estimates [in contrast to profile plots where p3, p4... take distinct values for each (p1,p2), maximizing the function for each (p1,p2)].
Logical: whether to show distinctly the points added in the latest iteration.
further arguments passed to or from other methods (currently can be used to pass a few arguments such as map.asp in all cases,
or variances to filled.mapMM).
The default color palette has been and may still be subject to changes, so if you want reproducible results, provide an explicit value. Some of the idiosyncratic color palettes used by (at least) early versions of Infusion helped me see what I want to see in a plot, but were not necessarily appropriate in all contexts. The current default for for “slice” plots in plot.SLik_j is
function(n) grDevices::hcl.colors(n=n, palette="viridis", rev=TRUE). The “turbo” palette as called by
function(n){ .viridisOpts(n=n, option="H", begin=0.1)} is being tried as default for filled, non-slice plots, and a more flashy one is used for dot plots.
Different graphic functions are called depending on the number of estimated parameters. For two parameters, mapMM or filled.mapMM are called. For more than two parameters, spaMM.filled.contour is called. See the documentation of these functions for the appropriate format of the plot.title and plot.axes arguments.
if (FALSE) {
## Using 'slik_j' object from the example in help("example_reftable")
plot(slik_j,filled=TRUE,
plot.title=quote(title("Summary-likelihood-ratio surface",
xlab=expression(mu),
ylab=expression(sigma^2))))
}
Run the code above in your browser using DataLab