PEMODELAN DATA KESEHATAN KABUPATEN BANYUWANGI DENGAN REGRESI TERBOBOTI GEOGRAFIS

  • Dinar Arga Prasetyo
  • Bambang Sumantri
  • Mohammad Masjkur

Abstract

Regression analysis is a method used to analyze
data and draw conclusions about the relationship between
dependence variables with response explanatory variables.
However modeling with ordinary regression is less precise
because in spatial case such as health case there is violation
of spatial heterogeneity. Geographically Weighted Regression
(GWR) is one of the point estimation which is effectively
solving problem with spatial heterogeneity violation. Generally
GWR brings ordinary regression framework modeling to a
weighted regression modeling with kernel gaussian weighting
function which is obtained locally model in each observation
locations. GWR modeling prove to produce model with Akaike
Information Criterion (AIC) less than AIC ordinary regression.
Whereas from F-Test Fotheringham, Brundson and Charlton
generate a conclusion that GWR is more effective to explain
a relationship between Health Index (Y) with Health Facilities
(X1), Population Density (X2) and Amount of Poority Peoples
(X3) with 5% significance level. From these results established
thematic maps as an evaluation of health programs so that a
reference to the improvement of the health sector.
Keywords-Regression, Geographically Weighted Regression,
Kernel Gaussian, Akaike Information Criterion.

Published
2013-04-01
Section
Articles