[R] Warning trying to plot -log(log(survival)) (John Sorkin)
therneau at mayo.edu
Tue May 12 15:00:49 CEST 2009
Nothing about the data set you posted should give issues in any version of the
package. The "points <=0" usually arises when the survival curve drops to zero,
and so log(-log(0)) is being trimmed off the graph by the plot command. This is
what should happen. As Thomas pointed out to you earlier, log-log survival
curves are not that useful and there are now much better methods.
I suspect that the data set you posted and the data set giving the warning
are not the same.
More information about the R-help