A distance-based model for spatial prediction using radial basis functions
- 69 Downloads
In the context of local interpolators, radial basis functions (RBFs) are known to reduce the computational time by using a subset of the data for prediction purposes. In this paper, we propose a new distance-based spatial RBFs method which allows modeling spatial continuous random variables. The trend is incorporated into a RBF according to a detrending procedure with mixed variables, among which we may have categorical variables. In order to evaluate the efficiency of the proposed method, a simulation study is carried out for a variety of practical scenarios for five distinct RBFs, incorporating principal coordinates. Finally, the proposed method is illustrated with an application of prediction of calcium concentration measured at a depth of 0–20 cm in Brazil, selecting the smoothing parameter by cross-validation.
KeywordsDetrending Distance-based methods Radial basis functions Random function models Smoothing parameter Spatial prediction
Work partially funded and supported by: Grant MTM2016-78917-R from the Spanish Ministry of Science and Education; Core Spatial Data Research (Faculty of Engineering, Francisco José de Caldas District University) (Grant COL0013969); and Applied Statistics in Experimental Research, Industry and Biotechnology (Universidad Nacional de Colombia) (Grant COL0004469).
- Capeche, C. L., Macedo, J. R., Manzatto, H. R. H., Silva, E. F.: Caracterização pedológica da fazenda Angra - PESAGRO/RIO, Technical report, Estação experimental de Campos (RJ). In Informação, globalização, uso do solo, Rio de Janeiro (1997)Google Scholar
- Cressie, N.: Geostatistics. The American Statistician 43, 197–202 (1989)Google Scholar
- Cuadras, C.M.: Distance analysis in discrimination and classification using both continuous and categorical variables. In: Dodge, Y. (ed.) Recent Developments in Statistical Data Analysis and Inference, pp. 459–474. Elsevier, North-Holland (1989)Google Scholar
- Cuadras, C.M.: Interpreting an inequality in multiple regression. The American Statistician 47(4), 256–258 (1993)Google Scholar
- Melo, C.E., Santacruz, A., Melo, O.O.: geospt: an R package for spatial statistics. R package version 1.0-2. http://geospt.r-forge.r-project.org/ (2012)
- R Development Core Team: R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org/ (2016)
- Thiébaux, H., Pedder, M.: Spatial Objective Analysis: With Applications in Atmospheric Science. Academic Press, London (1987)Google Scholar