Predicted fractional area of suitability is different to binary prediction based on MaxTSS threshold

[Louis]

I have used the "maximum training sensitivity plus specificity" threshold as the value to predict "suitability" vs unsuitability to create a binary map to assess extent of a species' range covered by protected areas.

According to the model results the area predicted to be "suitable" given the given corresponding threshold should be ~ 0.24 (24%) of the study area. However after using the corresponding threshold to create a binary map of suitability vs unsuitability the suitable area is only 15.10 % of the overall study area. I have changed several parameters and used a couple of different presence datasets and this seems to be consistent.

Am I missing something or simply misinterpreting the meaning of the term / value "Maximum training sensitivity plus specificity area"? [The same would happen if I used ESS]

[David Galbraith]

Were there cells with null values (coming from records with missing predictor data)?

--
You received this message because you are subscribed to the Google Groups "Maxent" group.
To unsubscribe from this group and stop receiving emails from it, send an email to maxent+un...@googlegroups.com.
To post to this group, send email to max...@googlegroups.com.
Visit this group at http://groups.google.com/group/maxent.
For more options, visit https://groups.google.com/d/optout.

[Louis]

No null records as far as I can see from the outputs and text summaries.

According to the model summary there were 3,940, 837 pixels with data and 3,938,563 "points with values for all grid cells" which corresponds exactly with the binary suitability raster that I created. No warnings for missing environmental data were generated during running any of the replicates.

["Ayşe S. Turak"]

Hi Louis,

Did you get an answer to this?

Ayse