[S] ltsreg for S+ 4

Statistik Mail (statsoft@stat.math.ethz.ch)
Tue, 10 Feb 1998 12:30:34 +0100


Dear All

We have a function who calls ltsreg. On S+ 3.4 for Unix everything works
ok, but on S+ 4(all Releases) for Windows we get annoying warning messages
even though the results seem to be ok.

More precisely:

S+ 3.4:
> test <- ltsreg(1:10,c(2,1,9,4,1,5,7,8,9,12),intercept=T)
--> no warning
> test2 <- ltsreg(cbind(1,1:10),c(2,1,9,4,1,5,7,8,9,12),intercept=F)
--> no warnings

S+ 4:
> test <- ltsreg(1:10,c(2,1,9,4,1,5,7,8,9,12),intercept=T)
--> no warning
> test2 <- ltsreg(cbind(1,1:10),c(2,1,9,4,1,5,7,8,9,12),intercept=F)
--> Warning:
The covariance matrix of the data is singular.
There are 10 observations in the entire dataset of
10 observations that lie on the line with equation
1 (x_i1-m_1)+ 0 (x_i2-m_2)=0
with (m_1,m_2) the mean of these observations.

Questions:
- Is this message in our example of any importance?
- Can the results still be trusted?

Thanks in advance
Marcel Wolbers
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news