MEL {hlr} | R Documentation |
Logistic regression using maximum estimated likelihood (MEL) in order to cope with separation issues.
MEL(x, y, delta = 0.01, epsilon = 1e-06, maxit = 100) ## S3 method for class 'MEL': print(x, ...) ## S3 method for class 'MEL': summary(object, ...) ## S3 method for class 'MEL': plot(x, ...)
x |
design matrix (n, p) for function MEL ; object of class
'MEL' for the print and plot methods |
y |
response vector |
delta |
constant |
epsilon |
precision constant |
maxit |
maximum number of iterations |
object |
object of class 'MEL' |
... |
further arguments to be passed to the methods |
The MEL
function fits the MEL-model to the data.
The print
method displays the model coefficients.
The summary
method displays the model coefficients and
displays the names of the components of the MEL output object.
The plot
function plots the observed response values against
the predicted values (using the MEL linear predictor) on link scale.
The actual logistic cdf is added as well.
Object of class 'MEL' with following components
MEL |
MEL estimates of the coefficients |
outMEL |
object of class 'glm' corresponding to the final fit |
Tobias Verbeke, largely based on original S-PLUS code by Peter J. Rousseeuw and Andreas Christmann
Rousseeuw, P.J. and Christmann, A. (2003). Robustness against separation and outliers in binary regression. Computational Statistics & Data Analysis, 43, 315 – 332.
Original S-PLUS code available at http://www.stoch.uni-bayreuth.de/de/CHRISTMANN
par(mfrow = c(2,1)) ### Example 1 for function MEL: data set has overlap set.seed(314) n <- 500 beta <- matrix(c(2), ncol=1) x <- matrix(rnorm(n), ncol=1) eta <- -3 + x y <- rbinom(nrow(x), 1, plogis(eta)) out <- MEL(x, y) print(out) summary(out) plot(out) title("overlap") ### Example 2 for function MEL: data set has no overlap y[eta <= -1] <- 0 y[eta > -1] <- 1 out <- MEL(x,y, delta=0.01) out plot(out) title("no overlap") ### Example 3 for function MEL: all responses are equal to 0 y <- rep(0, nrow(x)) MEL(x,y) ### Example 4 for function MEL: all responses are equal to 1 y <- rep(1, nrow(x)) out <- MEL(x,y) out