0

I calculated the AIC of a Maxent output from maxnet package in R. I expected a single value as in AIC for a stardanrd glm but I got a vector. I do not understand why and if it is possible to have a single value.

Example:

library(maxnet)

data("bradypus")

fit <- maxnet(p=bradypus$presence, data = bradypus[-1], 
              regmult = 2,  # to adjust regularization
              classes = "q")

library(ENMeval)

p.occs <- predict(fit, bradypus, clamp=T, type = "exponential")
ncoefs <- fit$betas

aic.maxnet <- aic.maxent(p.occs, ncoefs)
head(aic.maxnet)

      AICc delta.AICc      w.AIC
1 19565.07   7.056598 0.01451207
2 19565.14   7.123900 0.01403185
3 19565.11   7.095729 0.01423089
4 19565.11   7.095745 0.01423078
5 19565.11   7.095564 0.01423206
6 19565.11   7.095450 0.01423288

I would like to have a sigle value of AIC (or AICc), also because this is a single model.

Giordano
  • 81
  • 5
  • The second parameter `ncoefs` is the "number of non-zero model coefficients". You are passing the coefficients. – Roland Jan 13 '23 at 08:54

0 Answers0