Weighted least squares fit in parametric survival models
DESCRIPTION:
Weighted least squares fit in parametric survival models.
USAGE:
censorReg.wfit(x, y, w, method, parameter.fixed=NULL, ...)
REQUIRED ARGUMENTS:
x
the design matrix.
y
the response vector. This is a function of the response, the fit, and the
model derivatives.
w
the weight vector. This is a function of the observation weights and the
model derivatives.
method
method for the linear solver. "qr" is usual.
OPTIONAL ARGUMENTS:
parameter.fixed
A vector indicating the columns in x that are fixed.
...
VALUE:
list of values from the weighted least squares fit.
DETAILS:
This is almost exactly the same routine as
lm.wfit. The fact that the weights
in the raw model are actually frequencies while the weights in
censorReg.wfit can
be negative required some changes in the computations.
SEE ALSO:
EXAMPLES:
This is an internal routine not usually called directly by users.