Crossval function

[Håkon Fjeld]

Hi! 

I have a quick question regarding the crossval function in unmarked. When I am using the function on a fit list (models fitted with distsampfunction, with output density), are the MAE (mean absolute error) values on a log-scale? 

Thanks for any attention and answers :) 

Håkon 

[Ken Kellner]

Hi Håkon,

MAE here is calculated as the mean absolute value of the residuals. You can see the default function used for calculating statistics with 

unmarked:::RMSE_MAE

For a distsamp model with output density the residuals are calculated like:

y - (lambda * sampled area * p)

So, I think the answer to your question is no, they are not on the log scale.

Ken

[Håkon Fjeld]

Thanks for the Quik replay :)