cph fails with boolean input about rms

[harrelfe]

On 12/15/2015 03:03 PM, Max Gordon wrote:

It seems that |model.matrix()| returns TRUE together with the variable
name. Not sure when this happend, can't find any note of it
https://cran.r-project.org/src/base/NEWS but the following code used
to run fine:

set.seed(10)
n <- 500
ds <- data.frame(
ftime = rexp(n),
fstatus = sample(0:1,size = n,replace = TRUE),
x = factor(sample(LETTERS[1:4],size = n,replace = TRUE)),
boolean = sample(c(TRUE,FALSE),size = n,replace = TRUE),
subsetting = factor(sample(c(TRUE,FALSE),size = n,replace = TRUE)))

library(rms)
dd <<- datadist(ds)
options(datadist="dd")

fit <- cph(Surv(ftime,fstatus == 1)~ x + boolean,data=ds)

Now it throws an |Error in X[, mmcolnames, drop = FALSE] : subscript
out of bounds| as the mmcolnames has the name for the boolean variable
while the |X <- model.matrix(sformula, X)| returns the boolean column
with the name |booleanTRUE|.

Btw - congratulations on the second edition. Great to see that it has
finally arrived.

Thanks Max.

I hadn't tested recent changes with a logical variable. Will fix this
and will make a new release of rms soon as I've made other changes too.

Frank

—
Reply to this email directly or view it on GitHub
#15.

---

Frank E Harrell Jr Professor and Chairman School of Medicine

Department of *Biostatistics*   *Vanderbilt University*

[harrelfe]

This has now been fixed and I just submitted a new version (with the fix) to CRAN.