General Spatial Models (GSM) Approach on Baby Infant Mortality Data
| Main Authors: | Pramoedyo, Henny, Trisilia, Meilinda |
|---|---|
| Format: | Article info application/pdf eJournal |
| Bahasa: | eng |
| Terbitan: |
Natural B, Journal of Health and Environmental Sciences
, 2012
|
| Subjects: | |
| Online Access: |
https://natural-b.ub.ac.id/index.php/natural-b/article/view/151 https://natural-b.ub.ac.id/index.php/natural-b/article/view/151/135 |
Daftar Isi:
- Proximity and linkages between sites led to the emergence of the phenomenon of spatial linkages. Weighting matrix can be used to determine the proximity and linkages between spatial data or spatial relationships and can be used to calculate the coefficient of spatial dependencies. This study uses spatial panel data, namely the infant mortality rate (IMR) regional data taken from a unit area of development (SWP) Gerbangkertasusila and the Malang-Pasuruan SWP in the period 2005-2009. Those data use rook contiguity to make spatial weighting matrix. The aim of this study is to determine the model of what can be formed from the general spatial model (GSM) using panel data. Estimation of panel models with common effects approaches, fixed effects and random effects, will be followed by estimating the coefficient parameters of the general spatial model on panel data using maximum likelihood estimation method. From the prediction model by using software EViews7 note that all spatial panel data in this study followed the random effect model. To estimate the coefficient parameters of the general spatial model with Matlab-R2010 software is used to obtain a spatial lag / autoregressive models (SAR) of random effects and spatial error models (SEM) random effect. Model selection using the criteria as well as the largest R2 and corr2, and the smallest AIC values, the MSE values and the SC values. The best model for regional infant mortality data is the spatial error models (SEM) random effect.