Bayesian estimation of spatial temporal model with applications in air pollutant data
by Li Wai Ming
M.Phil. Information Systems, Business Statistics and Operations Management
ix, 36 pages : illustrations ; 30 cm
Analysis of geostatistics data is an important topic for policy makers and researchers.
Traditionally spatial and spatio-temporal model usually impose two major assumptions:
Gaussian random process for the observation and constant variance over time. In this
paper we relax these two assumptions by introducing a student t formulation with dynamic
variance and leptokurtosis over sptial location and time. An illustrative study on NO2
concentration in Hong Kong region displays differential variance behavior over time and
signal of heavy tail for several locations.
Permanent URL for this record: https://lbezone.hkust.edu.hk/bib/b1628029