Empirical Likelihoods
Construct the empirical log likelihood or empirical exponential family log likelihood for a mean.
EEF.profile(y, tmin = min(y) + 0.1, tmax = max(y) - 0.1, n.t = 25,
u = function(y, t) y - t)
EL.profile(y, tmin = min(y) + 0.1, tmax = max(y) - 0.1, n.t = 25,
u = function(y, t) y - t)y |
A vector or matrix of data |
tmin |
The minimum value of the range over which the
likelihood should be computed. This must be larger than
|
tmax |
The maximum value of the range over which the
likelihood should be computed. This must be smaller than
|
n.t |
The number of points between |
u |
A function of the data and the parameter. |
These functions calculate the log likelihood for a mean using either
an empirical likelihood or an empirical exponential family likelihood.
They are supplied as part of the package boot for demonstration
purposes with the practicals in chapter 10 of Davison and Hinkley (1997).
The functions are not intended for general use and are not supported
as part of the bootpackage. For more general and more robust
code to calculate empirical likelihoods see Professor A. B. Owen's
empirical likelihood home page at the URL
https://statweb.stanford.edu/~owen/empirical/.
A matrix with n.t rows. The first column contains the
values of the parameter used. The second column of the output
of EL.profile contains the values of the empirical
log likelihood. The second and third columns of the output of
EEF.profile contain two versions of the empirical
exponential family log-likelihood. The final column of the
output matrix contains the values of the Lagrange multiplier
used in the optimization procedure.
Angelo J. Canty
Davison, A. C. and Hinkley, D. V. (1997) Bootstrap Methods and Their Application. Cambridge University Press.
Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.