roblm {roblm} | R Documentation |
MM-regression estimators
roblm(formula, data = list(), weights, na.action, model = TRUE, x = FALSE, y = FALSE, singular.ok = TRUE, contrasts = NULL, offset = NULL, control = roblm.control())
formula |
formula |
data |
data |
weights |
weights |
na.action |
na.action |
model |
model |
x |
x |
y |
y |
singular.ok |
singular.ok |
contrasts |
contrasts |
offset |
offset |
control |
control |
This function computes and MM-regression estimator as described in Yohai (1987). It uses a bi-square re-desceding score function, and by default it returns a highly robust and highly efficient estimator (with 50% breakdown point and 95% asymptotic efficiency for normal errors). It uses an S-estimator (Rousseeuw and Yohai, 1984) for the errors scale. This estimator is also computed with a bi-square score function. This function uses the Fast-S algorithm of Salibian-Barrera and Yohai (2005) to compute the S-estimators.
Standard errors are computed using the stable
asymptotic formula of Croux, Dhaene and Hoorelbeke
(2004). There is also the option of using the
Robust Bootstrap of Salibian-Barrera and Zamar (2002).
See roblm.control
.
An object of class roblm
. A list that includes the
following components:
coef |
The MM-regression estimator |
scale |
The S-scale estimator |
s |
The auxiliary S-regression estimator |
cov |
The estimated covariance matrix of the regression coefficients |
residuals |
Residuals associated with the MM-estimator |
fitted.values |
Fitted values associated with the MM-estimator |
converged |
TRUE if the IRWLS iterations
converged |
Matias Salibian-Barrera
Rousseeuw and Yohai (1984); Yohai (1987); Salibian-Barrera and Yohai (2005)
data(coleman.dat, package='roblm') summary( roblm(y~., data=coleman.dat, control = roblm.control(compute.roboot=TRUE) ) )