[R-sig-Geo] potential BUG in BBMM

[andrewaduff]

R-sig-geo group, 

I have found similar problems with the BBMM package in R.  I ran my datasets
through brownian bridge movement model as implemented in Animal Space Use,
http://www.cnr.uidaho.edu/population_ecology/animal_space_use.htm and the R
package adehabitat,
http://cran.r-project.org/web/packages/adehabitat/index.html (in adehabitat
with the help of Tyler, the original poster).  Both of the other programs
ran fine with similar outputs and the only problems I encountered were
dealing with the various formats needed for time, projections needed for
coordinates, and offsetting of discretized (rounded) sample points. 

When I ran the sample dataset included with the BBMM package, some of the
time the sample dataset would preform properly and sometimes I would get a
result returned with NaN for all probability values on both R 2.11.1 32-bit
and 64-bit.  The author was helpful in troubleshooting, I just choose to
implement the brownian bridge in adehabitat because of Tylers success with
the package and my short timeline for troubleshooting.  The Author of BBMM
had no problems with the sample dataset on his machine.  I speculate that
either data issues were causing problems and were masked by the routines in
BBMM and these errors did not halt calculation.  The raptor datasets I
worked with had problems like data discretion (rounding errors introduced by
Argos sattelite limitations) but those were dealt with by offseting routines
written in SAS.  My point clouds were somwhat dense with 1000-4000 points
per bird per year.   

I would be interested in any experiences others had with the package as we
might be able to provide insight to the author address the problems
encountered in this posting. 

Andrew Duff 

-- 
View this message in context: http://r-sig-geo.2731867.n2.nabble.com/potential-BUG-in-BBMM-tp5289846p5844167.html
Sent from the R-sig-geo mailing list archive at Nabble.com.