Next: Issues in choosing a
Up: paciorek-gisspat
Previous: Goals of spatial smoothing
- kriging
- Gaussian process models (kriging a special case of this)
- penalized spline models
- any nonparametric regression method for two covariates with fully
flexible interaction
- e.g., neural network models
Choice of model often matters less than how smoothing parameter(s)
chosen
Chris Paciorek
2004-03-18