AUC, TSS and Binary Prob values for my models - any thoughts?

[Steve Research]

Hi,

I have been grinding my data over the last few weeks abd trying to eliminate any Bias.

I have tried the model without a bias file.
With a bias file generated using the R script to create a density map of other associated species.
A 'distance to roads' map.

Attached are the various AUC, TSS and Binary Propb stats for these models. Am I really getting a significant prediction improvement from using the bias files?

Your thoughts welcome.

Cheers,
Steve

[Steve Research]

Another thought has occurred to me is it OK to alverage the TSS values across the 5 replicates or should that just be expressed as a range of say 0.6490 - 0.7042?

Cheers
Steve