[R] Robust Regression - LTS

[Kylie-Anne Richards]

Hi,

I am using robust regression, i.e. model.robust<-ltsreg(MXD~ORR,data=DATA).
My question:- is there any way to determine the Robust Multiple R-Squared
(as returned in the summary output in splus)? I found an equivalent model in
the rrcov package which included R-square, residuals etc in it's list of
components, but when I used this package the only results returned were
equivalent to the LTS reg in the MASS package, which obviously indicates
that the summary method does not work for this class of models. 

If required:

##The output for the LTS reg (MASS) using print and summary

Call:
lqs.formula(formula = MXD ~ ORR, data = DATA, method = "lts")

Coefficients:
(Intercept)        ORR
  7.578e+08    2.533e+01

Scale estimates 1.333e+09 1.303e+09

              Length Class      Mode
crit             1   -none-     numeric
sing             1   -none-     character
coefficients     2   -none-     numeric
bestone          2   -none-     numeric
fitted.values 4899   -none-     numeric
residuals     4899   -none-     numeric
scale            2   -none-     numeric
terms            3   terms      call
call             4   -none-     call
xlevels          0   -none-     list
model            2   data.frame list



## The output for the LTS reg (rrcov) using print and summary
 
Coefficients:
Intercept      ORR
1.178e+09  2.387e+01

Scale estimate 1.722e+09

                 Length Class  Mode
best             2451   -none- numeric
raw.coefficients    2   -none- numeric
alpha               1   -none- numeric
quan                1   -none- numeric
raw.scale           1   -none- numeric
raw.resid        4899   -none- numeric
coefficients        2   -none- numeric
scale               1   -none- numeric
resid            4899   -none- numeric
lts.wt           4899   -none- numeric
crit                1   -none- numeric
rsquared            1   -none- numeric
residuals        4899   -none- numeric
intercept           1   -none- logical
method              1   -none- character
RD               4899   -none- numeric
X                9798   -none- numeric
Y                4899   -none- numeric
fitted.values    4899   -none- numeric

## The output for the LTS reg (SPLUS) using print and summary ****{What I am
wanting to achieve in R}****

> model.robust<-ltsreg(MXD~ORR,data=DATA,na.action=na.exclude)
> print(model.robust)
Method:
Least Trimmed Squares Robust Regression. 

Call:
ltsreg(formula = MXD ~ ORR, data = DATA, na.action = na.exclude)

Coefficients:
     Intercept         ORR 
 1.465502e+009 2.325200e+001

Scale estimate of residuals:  1639000000 

Total number of observations:  4899 

Number of observations that determine the LTS estimate:  4409 
> summary(model.robust)
Method:
[1] "Least Trimmed Squares Robust Regression."

Call:
ltsreg(formula = MXD ~ ORR, data = DATA, na.action = na.exclude)

Coefficients:
     Intercept         ORR 
 1.465502e+009 2.325200e+001

Scale estimate of residuals: 1639000000 

Robust Multiple R-Squared: 0.4733 

Total number of observations:  4899 

Number of observations that determine the LTS estimate:  4409 

Residuals:
          Min.       1st Qu.        Median       3rd Qu.          Max. 
 -228135629879   -1032103953    -231375637    1234533512   55539148696

Weights:
   0    1 
 588 4311




Thanks very much for any help you can offer. 

Kylie-Anne Richards