emGauss.default(object, subset, prior = <<see below>>, start = <<see below>>, control = emGauss.control()) emGauss.preGauss(object, prior = <<see below>>, start = <<see below>>, control = emGauss.control()) emGauss.missmodel(object, prior = <<see below>>, start = <<see below>>, control = emGauss.control())
emGauss.default
: a data frame or matrix containing the raw data.
When a data frame is
input, the model is specified by the numeric variables. When a matrix
is input, all variables are used.
emGauss.preGauss
: an object of class
"preGauss"
.
emGauss.missmodel
: an object of class
"missmodel"
containing
the results of a previous analysis. Any of the functions
mdGauss
,
completeGauss
,
emGauss
, or
daGauss
may be used to produce the
missmodel object.
object
is a data frame,
this expression may use variables in the data frame.
"priorGauss"
.
"ml"
(maximum likelihood),
"noninformative"
, and
"ridge"
(for the default ridge prior).
String matching is used, so the characters
"m"
,
"n"
, or
"r"
are
sufficient.
"priorGauss"
object is created by routine
priorGauss
.
"missmodel"
object is input, any value specified in a previous call
has priority over the default value (but not over any currently
specified value).
mdGauss
are the mean and variance--covariance matrix of a multivariate
normal distribution. Thus,
start
may be a list with vector
component
mu
giving the mean and matrix component
sigma
giving
the variance-covariance matrix. Alternatively, a class
"Gauss"
object created as the
paramIter
component of the class
"missmodel"
object may be used. Routines
mdGauss
,
daGauss
, and
emGauss
may be
used to create an appropriate
"missmodel"
object.
object
is
a class
"missmodel"
object. In this case argument
start
defaults
to the final estimates in the input
"missmodel"
object.
emGauss.missmodel
: If not given, argument
control
defaults to
the control parameters specified in the
call
statement of the input
"missmodel"
object, but only if these are of the correct class. If
these are not given (or are not of the correct class), then the
argument
control
defaults to
emGauss.control
values.
"missmodel"
is returned; see
for further details.
See the help file for for additional details.
emGauss.default(object = cholesterol) cholesterol.pre <- preGauss(data = cholesterol) emGauss.preGauss(cholesterol.pre) cholesterol.em <- emGauss(cholesterol) emGauss.missmodel(object = cholesterol.em) # start with last parameter estimates in cholesterol.em