Closed kenkellner closed 4 years ago
I like the idea of adding the cloglog link, but I'm hesitant about the getN
method for occu
fits. Statistically it's fine, but it ignores abundance-induced heterogeneity in p, so it seems inferior to using occuRN
. Can we hold off on getN
for now and just add the cloglog link?
Makes sense, I sent another PR without it.
Adds option to use complimentary log-log link function in
occu
with a new argumentlinkPsi="cloglog"
(defaults to logit). Also added a new helper functiongetN
which will calculate predicted abundance for these models.Based on Jonathan Cohen's post in the forums: https://groups.google.com/forum/#!searchin/unmarked/cohen%7Csort:date/unmarked/Xlp_QT7COJE/sLwRMomIAAAJ