Variable Selection of Heterogeneous Spatial Autoregressive Models via Double-Penalized Likelihood

Heteroscedasticity is often encountered in spatial-data analysis, so a new class of heterogeneous spatial autoregressive models introduced this paper, where the variance parameters are allowed to depend on some explanatory variables. Here, we interested problem parameter estimation and variable selection for both mean models. Then, unified procedure via double-penalized quasi-maximum likelihood proposed, simultaneously select important Under certain regular conditions, consistency oracle property resulting estimators established. Finally, simulation studies real data analysis Boston housing carried illustrate developed methodology.