enmSdmX 1.2.10 2024-12-07

trainGLM() now allows users to remove models that did not converge or had boundary problems. The function also works for factor predictors with > 2 levels. Thank you, A.L.!
trainNS() works when the number of predictors is >2 and interactions are allowed between variables. Thank you, P.T.! It also handles factors with >2 levels and allows users to remove models that did not converge or had boundary problems. Thank you, A.L.!

enmSdmX 1.1.9 2024-11-01

trainGAM() does not include interaction terms if interaction is NULL, and does not fail when interactions are included and the number of predictors is >2 (thank you, P.T.!).

Clarified mis-leading help in several functions, including all the trainXYZ() functions (thank you, PT!).

Replaced dependency on MuMIn with one one AICcmodavg for calculation of AICc. Received warning that MuMIn was going to be archived on CRAN.

Added function trainESM() for ensembles of small models.
Added several UTM coordinate reference systems accessible through getCRS().
Fixed bug in predictEnmSdm() for predicting kernel density models from the ks package.

trainGLM(), trainNS(), and predictEnmSdm() now have options to automatically center and scale predictors. If a GLM or NS model created using trainGLM() or trainSN() is provided to predictEnmSdm(), it will automatically scale the predictors properly.

Removed dependency on dismo, replaced where possible by predicts; copied gbm.step() and predict() method for MaxEnt to enmSdmX as a momentary fix; would love a professional solution!
Added European Datum 1950 (ED50) to coordinate reference systems table
Added ability to export intermediate values for plotting in evalContBoyce()
Fixed spelling errors in crss data table (coordinate reference system names)

Backwards incompatible: trainRF() uses ranger package for random forests (changed from randomForest package)
Backwards incompatible: geoFold() uses complete clustering method by default (changed from single)
New feature: trainGLM() and trainNS() can automatically scale predictors
New feature: Reworked trainGLM() to handle cases with large numbers of predictors
New feature: Added geoFoldContrast() for assigning geo-folds to background or absence sites
Better functionality: trainRF() now indicates if the response is binary using a binary argument (vs. family, which falsely implied more functionality than RFs have)

Better functionality: Expanded capacity of .calcWeights() (a hidden function) to handle different values for family

new feature: Added help page for troubleshooting running functions that support parallel operation
Bug fix: predictEnmSdm() when using a BRT model and predicting to a SpatRaster (thank you, Nikki C!)

New feature: Users can select clustering method in geoThin() and geoFold() (thank you, Pascal T!)
Bug fix: Example in customCRS() when GADM server is down
Bug fix:geoThin() which returned input if it was a data.frame*
Bug fix:geoThin() which returned a data.frame lacking coordinates if input was a data.frame (thank you, Pascal T!)

New feature: getCRS() (no arguments) now displays shiny table of all available CRSs
Bug fix: Fixed issue in nearestGeogPoints() when polygon lay under centroid
Bug fix: Fixed bug with “table” call of getCRS()

New feature: summarizeByCrossValid() now summarizes natural spline (NS) models
Bug fix: Bug experienced by some users using predictEnmSdm() and predictMaxNet() (Thank you, Nikki!)
Bug fix: trainByCrossValid() using improper call to evalContBoyce()
Bug fix: extract() bug in some examples