Someone just sent me a data set that causes the lmekin function, part of
the kinship package, to fail. In chasing it down I get an error I have
never seen before.
fit <- lmekin(icam1 ~ factor(center) + age + factor(sex),
random= ~1|iid, data=chaidata, varlist=kmat)
Error in Y - fitted : non-numeric argument to binary operator
Add the recover option, and the offending lines are
fitted <- c(X %*% lfit$coef) #fitted, on the original scale
residuals <- Y - fitted
> options(error=recover)
> fit <- lmekin(icam1 ~ factor(center) + age + factor(sex),
+ random= ~1|iid, data=chaidata, varlist=kmat)
Error in Y - fitted : non-numeric argument to binary operator
Enter a frame number, or 0 to exit
1: lmekin(icam1 ~ factor(center) + age + factor(sex), random = ~1 | iid,
data
Selection:
Enter an item from the menu, or 0 to exit
Selection: 1
Called from: eval(expr, envir, enclos)
Browse[1]> dim(X)
[1] 2601 6
Browse[1]> lfit$coef
(Intercept) factor(center)2 factor(center)3 factor(center)4
age
217.9110997 -2.9079576 -0.2147915 -7.8141818
0.5210394
factor(sex)2
5.1384741
Browse[1]> X %*% lfit$coef
`__Deferred_Default_Marker__`
----------------------------------------------
X is a model.matrix. If I save the two objects and reload them into a
fresh R session the multiplication works just fine.
I haven't modified this code in several years, and it is widely used.
All the C code is shared with coxme. Subsetting the data, or modifing
the fixed effects has no impact on the error. The same code runs in
Splus. At the point of error the computations have finished and it is
putting together a result list: all of the results are correct.
Nevertheless, my best (only) guess is memory corruption.
Questions:
Is this a rational guess? Any others?
Can you give any further insight on the error message?
Terry Therneau
sessionInfo()
R version 2.10.0 (2009-10-26)
x86_64-unknown-linux-gnu
locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_US.UTF-8 LC_COLLATE=C
[5] LC_MONETARY=C LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=en_US.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
attached base packages:
[1] grDevices datasets splines graphics utils stats
methods
[8] base
other attached packages:
[1] survival_2.35-9 rlocal_1.5.3
>
______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
