Model evaluation metrics for the 36 MaxEnt models for each of the 103 species

Evaluation metrics of the models for each species.
Explanation of the columns:
rm= regulization multiplier
fc=feature class. L stands for linear, H for hinge, Q for quadratic, LQ for linear-quadratic, LQH linear-quadratic-hinge and LQHP for linear-quadratic-hinge-product features
tune.args= the combination of rm and fc
auc.train= Average AUC (area under the ROC curve) for the training data across partitions
cbi.train= Average Boyce index for the training data across partitions
auc.diff.avg= Average difference between training and validation AUC (measure of overfitting)
auc.diff.sd= Standard deviation of that difference
auc.val.avg= Average validation AUC across partitions
auc.val.sd= Standard deviation of validation AUC across partitions
cbi.val.avg= Average validation Boyce Index across partitions
cbi.val.sd= Standard deviation of validation Boyce Index across partitions
or.10p.avg= Average omission rate at the 10th percentile training presence threshold
or.10p.sd= SD of that omission rate
or.mtp.avg= Average omission rate at the minimum training presence threshold
or.mtp.sd=SD of that omission rate
AICc=Akaike Information Criterion corrected for small sample size
delta.AICc= Difference between the AICc of this model and the best model
w.AIC=Akaike weight = relative likelihood of the model being the best among those compared.
ncoef= number of coefficients in the MaxEnt model