error in val.prob function - group=T

[Carlos Davidson <cdavidson@csus.edu>]

Dear S list

When I run val.prob with the group=T option I get a warning message
saying "NAs generated in: log(x)". When I do not use the
group=T option I do not get an error. Otherwise the output looks fine.
Below is code and output that generates the warning message. fits2 is the
"b" array of probabilities, and ramumain01 is the "y"
array of 0 or 1 values. When I I thought the problem might be that my
probabilities go down to near zero, but see code and output below of an
example where the minimum probability is 0.12 and I still get the same
warning message. I can run the example in the val.prob documentation
without any problems. And I can adjust the example sample size to match
mine and run either the example b array with my y array or vice versa
without problems - only when I use both my b and y arrays do I get an
error. And I get this same error with two completely different datasets!
Any suggestions? 

> summary(fits2)
      Min.   1st Qu.   
Median      Mean   3rd
Qu.      Max. 
 0.0000000 0.0203278 0.0584008 0.1327387 0.1728797 0.8852888
> summary(ramumain01)
   Min. 1st Qu. Median   Mean 3rd Qu.  
Max. 
 0.0000 0.0000  0.0000 0.1327 0.0000  1.0000
> valobj <- val.prob(fits2, ramumain01)
> valobj <- val.prob(fits2, ramumain01, group = T)
Warning messages:
  NAs generated in: log(x)



# here with manipulated fits values so none near zero.

> fits3 <- (fits2 + 0.2) * 0.6
> summary(fits3)
      Min.   1st Qu.   
Median      Mean   3rd
Qu.      Max. 
 0.1200000 0.1321967 0.1550405 0.1996432 0.2237278 0.6511733
> valobj <- val.prob(fits2, ramumain01, group = T)
Warning messages:
  NAs generated in: log(x)

Thanks!!

Carlos 

*************************** 
 
Carlos Davidson 
 
Department of Environmental Studies 
California State University, Sacramento 
6000 J Street 
Sacramento, CA 95819-6001 
 
Phone: (916) 278-6063 
Fax: (916) 278-7582 
Email: cdavidson@csus.edu 
 
Web:
www.csus.edu/indiv/d/davidsonc