Data and scripts for "Evaluating the predictive performance of presence-absence models: why can the same model appear excellent or poor?"

These files contain the scripts and data for reproducing the results of the article "Evaluating the predictive performance of presence-absence models: why can the same model appear excellent or poor?".

These scrips compute four metrics (AUC, Tjurs R2, max-Kappa and max-TSS)  measuring the predictive performance of presence absence models applied to a fungal dataset (Fungi_data.csv). To compare the metrics, these are applied at different spatial hierarchical scales and using different cross validation strategies.

The scrips should be used in R and in the following order: (1) S1 Define models, (2) S2_1 Fit models and evaluates AUC TjurR2, (3) S2_2 Add Kappa tss, (4) S2_3 Average replicates, (5) S3 Show examples Fig. 1, (6) S4 Show relationships among measures, (7) S5 Show hierarchical results Fig.3.