Home > Model selection for spatially correlated data

Model selection for spatially correlated data


Resumen Paper






Andrés Riquelme1.

1.Universidad de Talca,        juriquelme@utalca.cl.



 The aim of this paper is to compare several model selection techniques in detren- ding the error process under spatially dependent data with correlated regressors. Among the available model selection methods I compare the Adaptive Lasso (alasso) estimator (Zou, 2006), the smoothly clipped absolute deviation (scad) estimator (Fan and Li, 2001), the least absolute shrinkage and selection (las- so) operator (Tibshirani, 1996) and elastic net estimator (Zou and Hastie, 2005). Among the various error processes I focus in fitting the Exponential covariance functions for the variogram  estimation.

I found that the adaptive lasso is the best among the analyzed techniques.

Keywords: Shrinkage; Spatially dependent data; lasso variogram.


Participación en sesión: Métodos y modelos de análisis regional.