ABSTRACT
For spatial predictive modeling, the ultimate purpose is to generate spatial predictions by developing and applying an accurate predictive model. Thus, the predictive accuracy of the model is critical. This chapter defines the concepts of observed and predicted values and associated errors, explores the relationships among various errors, and introduces various accuracy and error measures that can be used to evaluate predictive models for numerical and categorical data. The function pred.acc and vecv in the spm package is introduced for the recommended and commonly used accuracy and/or error measures. The function tovecv in the spm package is also introduced for converting some error into VEcv. It also introduces various model validation methods, and techniques to handle randomness associated with cross-validation methods, with reproducible examples in R .
