When testing simulated species distributions against field observations, two issues are of primary interest: (1) the optimized cut-off level to predict "presence" given the simulated probabilities of occurrence, and (2) the accuracy of the statistical model for this optimized (and other) cut-off level(s). The accuracy can be assessed through a number of different measures. The optimization of the cut-off level is then dependent on the statistics chosen. Therefore, SimTest calculates different measures, including their respective optimized cut-off levels.

SimTest allows to calculate a series of test measures such as Kappa statistics, AUC-statistics, omission or commission errors, and it allows to calculate optimized cut- off levels to classify simulated probabilities (of occurrence) into presence and absence. Additionally, SimTest produces a series of output files that contain necessary data to further analyze or illustrate (e.g. ROC-plot) the data. The most important measures are directly presented on the terminal window (click here to see an example).

You can dowmload a manual to get more detailed instructions and some theory behind the calculations. Simtest reads a file called simtest.dat to read the input. You can download an example file here.

General specifications of the AML:

References:

 

[top] [back] [home]