Description Usage Arguments Details Value Note
Estimators for the PIM parameters
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 | estimator.nleqslv(jac = NULL, method = c("Broyden", "Newton"),
global = c("dbldog", "pwldog", "qline", "gline", "none"),
xscalm = c("fixed", "auto"), control = list(ftol = 1e-06),
scoreFunctionCreator = scorefunctioncreator.default,
treat.convergence.error = c("warn", "error", "log", "ignore"))
estimator.glm(control = list())
estimator.glmnet(alpha = 1, nlambda = 100, lambda = NULL,
standardize = TRUE, penalize.intercepts = FALSE)
estimator.BB(method = c(2, 3, 1), control = list(), quiet = FALSE,
scoreFunctionCreator = scorefunctioncreator.default,
treat.convergence.error = c("warn", "error", "log", "ignore"))
estimator.trymultiple(scoreFunctionCreator = scorefunctioncreator.default)
scorefunctioncreator.default(Z, Y, link, W = NULL)
estimator.lqa(control = lqa.control(), penalty = NULL,
method = "lqa.update2", standardize = TRUE, penalize.intercepts = FALSE)
|
jac, global, xscalm |
See |
method |
See |
control |
See |
scoreFunctionCreator |
Function that will create the score vector function. This
defaults to |
treat.convergence.error |
Defaults to |
alpha, nlambda, lambda, standardize |
See |
penalize.intercepts |
If an intercept is present in the model, penalize it or not. |
quiet |
See |
Z |
Pseudo-observation design matrix |
Y |
Pseudo-outcomes. |
link |
Name of the link function. |
W |
Weights to be applied to the pseudo-observations. |
penalty |
Any |
These functions estimate the coefficients (/parameters) of the PIM by solving the (sometimes) nonlinear equations. Each calls upon another library for the solution and some may be more accurate / correct / performant depending on the specific model.
estimator.glm
returns the regular glm
estimate (assuming the pseudo-
observations are independent)
estimator.glmnet
returns the elsatic net penalized glmnet
estimate
(assuming, again, the pseudo-observations are independent)
For the different implementations, morefitinfo
contains:
estimator.nleqslv
The return value of the nleqslv
call
estimator.glm
The return value of the glm
call
estimator.glmnet The return value of the glmnet
call, with some added
items: usedalpha
, usedfamily
,usedoffset
and standardize
estimator.BB
The return value of the BBsolve
call
estimator.trymultiple
is a special case that will simply try to use
nleqslv
, and with several parameters, hoping that one may
lead to a fit. Although it provides the best chance of getting a fit, take care in
assuring that the fit matches the covariance estimator! Also be aware that this may
be slow.
These functions (estimator.*
) each return a function themselves. The returned
function should have three parameters (startvalues
, a set of initial estimates of the
parameters; pfd
, an object of class pimfitdata
; link
, the name
of the link function) and should itself return a list of two items:
coefficients |
The parameter estimates |
morefitinfo |
Implementation specific information on the fit |
For scorefunctioncreator.default
: a function that takes a set of parameter
estimates (beta
) and calculates a set of values that should be solved for zero.
For estimator.glmnet
, coefficients
contains a sparse matrix
holding the coefficient estimates and intercept (!) for all lambda values.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.