search

DistanceDevelopment / mrds

R package for mark-recapture-distance-sampling analysis
GNU General Public License v3.0
4 stars 4 forks source link

Hazard rate optimisation with bins #72

Open erex opened 1 year ago

erex commented 1 year ago

Prompted by this user question regarding failure of mrds 2.2.8 to fit a hazard rate to binned data with error

Error in integrate: the integral is probably divergent. All models failed to fit.

while mrds 2.1.18 (circa 2018) was able to fit the hazard rate data to the same model. The natural question is "What does DistWin do?"

The answer is that DistWin is successfully able to fit said model to said data without the error message. Note, the binning with DistWin was done with cutpoints 0, 99, 199, 299, 399 to align with how ds bins when argument cutpoints=c(0,100,200,300,400).

Hence mrds 2.2.8 falls farther from DistWin hazard rate optimisation (for this data set) than did mrds 2.1.18

Data provided by user and available (along with DistWin project) upon request

LHMarshall commented 1 year ago

@erex let's discuss this at next dist dev

lenthomas commented 11 months ago

@LHMarshall can you please contact the original user and let the know about the new mcds.exe optimizer, and see if that fixes the issue for now (while we do further work on the optimizer).