user chooses the grid scale -A- select the species which have adequate sample sizes -A- scalingMetric is median, quartile or random -A- for each species: clipping generates min conv. poly around occurance pts, then buffers by n times the area of the poly. then clip the envlayers to the poly -A- DiGIR returns all species occurance -A- we need to iterate each species -A- the structure of the dataset is that each row (xml tagged) is a dv followed by ivs. all values must be a single byte there is a specific xml format of input (internal representation) DV=(0|1) default with ENM | (0-254) (255 indicates a value > 254) IVs=1-254 -A- 1-254 -A- could be repurposed as general binning util -A- metadata for rescaling: this can be applied to an envlayer and then parameterize input west 2 of garpPrediction actor -A- select from the following: -medians (systematic) -quartiles (systematic) -quartiles (random) -infinite (random) -manual -A- the env layer MD determines the allowed vals of scaleValue -A- if the species occ count >= threshold then continue else return -A- Presample layer files for use with the GarpAlgorithm Actor. Create a prediction based on the output of a GarpAlgorithm actor. Run the GARP algorithm when the loop is still running, it passes origInput through origOutput. when it is done, it passes true out the done output -A- -save the iteration matrix -select the runs that meet the threshhold of min ommission (params in DTGarp (middle panel)) -calc the medium commission of the min ommission models -select the user-defined percentage of the min ommission models that are closest to the median commission (from prev step) -A- input1: species selection prior to sampling input2: output of clipping -A- for each climate change layerset (including no change layersets aka origEnvLayers) for each ruleset in selected rule matrix get garp prediction on layerset end for sum grids end for -A- for each grid in gridSet apply grid mask end for -A- for each set run ROC algorithm end for -A- ROC = Reciever operating characteristic insert terry's phd student's algorithm here -A- this generates one output grid for each input grid -A- ***need to apply the threshold generator to the orig present dataset this produces a threshold that we use to reclassify the grids -A- For each grid summarize frequency distro of cells end for for each pair-wise grid combo of ((present or present-reclass) ^ (climate or climate-reclass)) summarize frequency distro of cells (on the intersections) end for for each pair-wise grid combo of ((present or present-reclass) v (climate or climate-reclass)) conditioned on contiguity of climate cells with present cells when both are predicted present) for union outcome present presence gets flagged and climate cells with no flagged neighbor are set to not present end for For each grid (in outcome of loop above) summarize frequency distro of cells end for -A-