lms
They represent a robust fit of a linear model.
This class of objects is returned from the
lmsreg
function.
The
"lms"
class of objects has methods for the following generic
functions:
plot
,
print
,
summary
.
The following components must be included in a legitimate
lms
object.
quan
th-order statistic of the absolute value of the residuals
over the
quan
th-order statistic of the absolute value
of y - theta, where theta is minimized over the denominator.
Here
quan
is the order statistic to be minimized,
and theta is not subtracted from y if
intercept = FALSE
.
quan
th-order statistic of each column of
the absolute value of the residuals, where
quan
is the order statistic that is being minimized.
y
containing the residuals from the regression.
Rows corresponding to missing values in
x
or
y
have
NA
s.
y
of weights for use in
weighted least squares.
These weights are
1
for points with reasonably small residuals and are
0
for points with large residuals.
This is only included if
wt
is TRUE in the call to
lmsreg
.
intercept
.
y
of resistant diagnostics,
For each trial the absolute value of the residuals is divided by the objective
for that trial;
the maximum of these numbers is retained for each observation.
The
diagnostic
component consists of these maxima divided by their median
for each response variable.
This is only included if
diagnostic
is
TRUE
in the call to
lmsreg
.
If
diagnostic=FALSE
, then computation as well as space is saved.
mve=TRUE
),
the robust distances of the cases obtained from
cov.mve
are returned.
lmsreg.default
or the data matrix.
model=TRUE
(in
lmsreg.formula
).
x=TRUE
(in
lmsreg.formula
).
y=TRUE
(in
lmsreg.formula
).