Displays a series of plots of the profile t function and interpolated
confidence intervals for the parameters in a nonlinear regression
model that has been fit with nls and profiled with
profile.nls.
# S3 method for profile.nls
plot(x, levels, conf = c(99, 95, 90, 80, 50)/100,
absVal = TRUE, ylab = NULL, lty = 2, …)an object of class "profile.nls"
levels, on the scale of the absolute value of a t
statistic, at which to interpolate intervals. Usually conf
is used instead of giving levels explicitly.
a numeric vector of confidence levels for profile-based
confidence intervals on the parameters.
Defaults to c(0.99, 0.95, 0.90, 0.80, 0.50).
a logical value indicating whether or not the plots
should be on the scale of the absolute value of the profile t.
Defaults to TRUE.
the line type to be used for axis and dropped lines.
other arguments to the plot.default
function can be passed here (but not xlab, xlim,
ylim nor type).
The plots are produced in a set of hard-coded colours, but as these
are coded by number their effect can be changed by setting the
palette. Colour 1 is used for the axes and 4 for the
profile itself. Colours 3 and 6 are used for the axis line at zero and
the horizontal/vertical lines dropping to the axes.
Bates, D.M. and Watts, D.G. (1988), Nonlinear Regression Analysis and Its Applications, Wiley (chapter 6)
# NOT RUN {
require(graphics)
# obtain the fitted object
fm1 <- nls(demand ~ SSasympOrig(Time, A, lrc), data = BOD)
# get the profile for the fitted model
pr1 <- profile(fm1, alpha = 0.05)
opar <- par(mfrow = c(2,2), oma = c(1.1, 0, 1.1, 0), las = 1)
plot(pr1, conf = c(95, 90, 80, 50)/100)
plot(pr1, conf = c(95, 90, 80, 50)/100, absVal = FALSE)
mtext("Confidence intervals based on the profile sum of squares",
side = 3, outer = TRUE)
mtext("BOD data - confidence levels of 50%, 80%, 90% and 95%",
side = 1, outer = TRUE)
par(opar)
# }
Run the code above in your browser using DataLab