NA AIC in new version of ENMeval

40 views
Skip to first unread message

fateme...@gmail.com

unread,
Jul 27, 2021, 4:07:43 AMJul 27
to Maxent
Hello everyone,
I used ENMeval for running Maxent for my post-doc work using following code:

eval1<- ENMeval::ENMevaluate(occ = occs [, c("long", "lat")], 
                              env = allstack, tune.args = list(fc = c("L", "H", "Q", "P", "LQ", "LQH","LQHPT"), rm = 1:5),   partitions = "block", n.bg = 10000,  parallel = FALSE, algorithm = 'maxent.jar', 
                              user.eval = proc, doClamp = TRUE, bin.output = TRUE)

, but I received NA for some AIC value models. Also, in the new version, we can not set rasterPreds = TRUE to solve it. I really appreciate it if anyone could help me to fix this issue?

Best Regards,
Fatemeh Moein


Jamie M. Kass

unread,
Sep 15, 2021, 7:13:56 AMSep 15
to Maxent
You can get NA for AICc for some models if there are more predictors than data points. This can happen when using complex feature classes like hinge or product with small datasets. AIC cannot be calculated if the number of predictors exceeds the number of training data points.

Jamie
Reply all
Reply to author
Forward
0 new messages