Modeling Method Summary

Below is a list of the methods we have examined so far.

	Response	Covariates/Predictors	Model Parameters	Visualization	Performance Metrics
Linear Regression	Continuous	Continuous	None	Response plots	R2, residual distribution
General Linear Regression: log transform	Continuous	Continuous	Type of transformation (e.g. log, exponential)	Response curves	R2, residual distribution
Generalized Linear Regression: binomial	Presence/Absence	Continuous		Response curves	Residual deviance, residual distribution
Generalized Linear Regression: poisson	Counts	Continuous?		Response curves	Residual deviance, residual distribution
Generalized Additive Method (GAM)	Continuous	Continuous/Categorical	gamma	Response curves	AIC, residual distribution
Categorical Trees	Categorical	Continuous/Categorical	cp, minbucket, minsplit	Decision tree	xerror (test error rate), number of nodes, sensitivity & specificity or confusion matrix
Regression Trees	Continuous	Continuous/Categorical	cp, minbucket, minsplit	Decision tree	xerror (test error rate), residual distribution
MaxEnt	Occurrences	Continuous/Categorical	Regularization	Response curves	AUC, AIC available in ENMTools and BlueSpray
HEMI 2	Occurrences	Continuous/Categorical	Locations of control points	Response curves	AIC, AUC, others?

Residual distribution refers to examining how the residuals are distributed using histograms, means, standard deviations, QQPlots and other tools.

Geospatial Activities