ABSTRACT
This chapter introduces modern statistical methods, which are also gradient based/detrended spatial predictive methods. Modern statistical methods use secondary information or predictive variables for detrended spatial predictive modeling. In this chapter, the following modern statistical methods that are available in R are introduced: 1) linear models; 2) trend surface analysis; 3) thin plate splines; 4) generalized linear models; and 5) generalized least squares. Various examples are provided for how to apply these methods. In this chapter it is also demonstrated why models/variables selections should be based on predictive accuracy. The function glmcv , glmnetcv , tpscv and glscv in the spm2 package are introduced for developing optimal predictive models for these methods. And the relevant functions are introduced for generating the spatial predictions of these methods. The petrel point data and petrel.grid grid data in the spm package are used to demonstrate the applications of the modern statistical methods.
