feglm Control ParametersSet and change parameters used for fitting feglm.
Note: feglm.control is deprecated and will be removed soon.
feglmControl(dev.tol = 1e-08, center.tol = 1e-05, rho.tol = 1e-04,
conv.tol = 1e-06, iter.max = 100L, limit = 10L, trace = FALSE,
drop.pc = TRUE, pseudo.tol = NULL, step.tol = NULL)feglm.control(...)
tolerance level for the first stopping condition of the maximization routine. The
stopping condition is based on the relative change of the deviance in iteration \(r\)
and can be expressed as follows: \((dev_{r - 1} - dev_{r}) / (0.1 + dev_{r}) <
tol\). Default is 1.0e-08.
tolerance level for the stopping condition of the centering algorithm.
The stopping condition is based on the relative change of euclidean norm in iteration \(i\) and
can be expressed as follows: \(||\mathbf{v}_{i} - \mathbf{v}_{i - 1}||_{2} <
tol ||\mathbf{v}_{i - 1}||\). Default is
1.0e-05.
tolerance level for the stephalving in the maximization routine. Stephalving only takes
place if the deviance in iteration \(r\) is larger than the one of the previous iteration. If
this is the case,
\(||\boldsymbol{\beta}_{r} - \boldsymbol{\beta}_{r - 1}||_{2}\) is
halfed until the deviance is less or numerically equal compared to the deviance of the previous
iteration. Stephalving fails if the the following condition holds: \(\rho < tol\),
where \(\rho\) is the stepcorrection factor. If stephalving fails the maximization
routine is canceled. Default is 1.0e-04.
tolerance level that accounts for rounding errors inside the stephalving routine when
comparing the deviance with the one of the previous iteration. Default is 1.0e-06.
unsigned integer indicating the maximum number of iterations in the maximization
routine. Default is 100L.
unsigned integer indicating the maximum number of iterations of
theta.ml. Default is 10L.
logical indicating if output should be produced in each iteration. Default is FALSE.
logical indicating to drop observations that are perfectly classified
(perfectly seperated) and hence do not contribute to the log-likelihood. This option is useful to
reduce the computational costs of the maximization problem, since it reduces the number of
observations and does not affect the estimates. Default is TRUE.
deprecated; use center.tol instead.
depreacted; termination conditions is now similar to glm.
arguments passed to the deprecated function feglm.control.
The function feglmControl returns a named list of control
parameters.