Semiparametric Spatial Single Index Model

• Activity Number: 174
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Hamdy Fayez Farahat Mahmoud*+ and Inyoung Kim
• Companies: Virginia Tech and Virginia Tech
• Keywords: Markov Chain Expectation Maximization ; Spatially correlated data ; Single index model ; Semiparametric regression ; Mixed model

Abstract

In this paper, we propose two semiparametric single index models for spatially correlated data. In the first model, the semiparametric function and the spatially correlated random effects are additively separated. The second model considered does not separate the semiparametric function and spatially correlated random effects. We estimate these two models using two different algorithms, both based on Markov Chain Expectation Maximization (MCEM) algorithm. Our proposed models are compared using simulations, which suggest that the semiparametric single index nonadditive model provides more accurate estimates of spatial correlation. Both methods are demonstrated using mortality data collected from six cities in Korea.