Bootstrapping for nonlinear models
Bootstraping for nonlinear models
boot_nls(
object,
f = NULL,
R = 999,
psim = 2,
resid.type = c("resample", "normal", "wild"),
...
)object |
object of class |
f |
function to be applied (and bootstrapped), default coef |
R |
number of bootstrap samples, default 999 |
psim |
simulation level for |
resid.type |
either “resample”, “normal” or “wild”. |
... |
additional arguments to be passed to function |
The residuals can either be generated by resampling with replacement
(default or non-parametric), from a normal distribution (parameteric) or by changing
their signs (wild). This last one is called “wild bootstrap”.
There is more information in boot_lm.
require(car) data(barley, package = "nlraa") ## Fit a linear-plateau fit.nls <- nls(yield ~ SSlinp(NF, a, b, xs), data = barley) ## Bootstrap coefficients by default ## Keeping R small for simplicity, increase R for a more realistic use fit.nls.bt <- boot_nls(fit.nls, R = 1e2) ## Compute confidence intervals confint(fit.nls.bt, type = "perc") ## Visualize hist(fit.nls.bt, 1, ci = "perc", main = "Intercept") hist(fit.nls.bt, 2, ci = "perc", main = "linear term") hist(fit.nls.bt, 3, ci = "perc", main = "xs break-point term")
Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.