drgee (version 1.0.1)

geeFit: Generalized Estimating Equation Solver

Description

geeFit is basically a wrapper around glm.fit but also computes the residuals and their derivatives with respect to the estimated coefficients.

Usage

geeFit(y, x, link=c("identity","log","logit"), ...)

Arguments

y

The outcome observations.

x

The design matrix.

link

A character string naming the link function. Have to be "identity", "log" or "logit".

...

Further arguments to be passed to glm.fit

Value

• A list with the following elements:
• coefficients
  A named vector of the estimated coefficients.
• fitted.values
  A vector of the predicted outcomes evaluated with the estimated coefficients.
• res
  A vector of the residuals, i.e. the difference between the the observed outcome y and fitted.values.
• d.res
  The derivatives of the residuals with respect to the model parameters evaluated with the estimated coefficients. A matrix with rows corresponding to observations and columns corresponding the the estimated coefficients.

encoding

latin1

Details

geeFit estimates coefficients $\theta$ using an estimating equation $W{R-k^{-1}(\theta^T W)}=0$ where $k$ is the identity, log or logit link function. glm.fit is used to estimate the coefficients. The family object used in the glm.fit call depends on the argument link. It will be gaussian(), quasipoisson() or binomial() when link is "identity", "log" and "logit" respectively.

geeFit is intended to be used as help function in drFit, ebeFit and obeFit.

See Also

glm.fit