A great opportunity on the availability of geo-reference data allows researcher to analyze not only on its spatial pattern but also including space-time interactions among regions. Spatio-temporal models designed with hierarchical fashion will be presented in this research proposal, with INLA (Integrated Nested Laplace Approximation) as the current estimation method for Bayesian analysis. INLA based on latent Gaussian posterior distribution which provides great computational benefit rather than Markov Chain Monte Carlo (MCMC). We will model the poverty data set using classical, dynamic and space-time interaction of spatio-temporal models, and investigate the poverty relationship with socio-economics information on its ecological regressions. All computational aspect will be solved using R-INLA package and deviance information criteria for models best fit selection

Areal Data, Gaussian Markov Random Field, Integrated Nested Laplace Approximation, Markov Chain Monte Carlo, Random Walk

• Drs. Iwa Sungkawa, M.S.

• Dr. Anita Rahayu, S.Si., M.Si.

• Ro'fah Nur Rachmawati, S.Si., M.Si