Estimation of Percentage on Malnutrition Occurrences in East Java using Geographically Weighted Regression Model
Abstract: The Province of East
Java has its own characteristics that differentiate it from any other regions.
Dissimilarities in characteristics of a region may encompass issues such as
social, economic, cultural, parenting, education, and the environment, so as to
cause the difference in case of severe under nutrition between one region to
another. Sufferers of malnutrition in one region may be linked and influenced
by the surrounding regions. Therefore, we need a statistical modeling that is
able to take into account the spatial factor. Statistical methods that can be
used to analyze the data and also takes into account the spatial factor are the
Geographically Weighted Regression (GWR). This study is aimed to determine the
case of malnutrition models in East Java Province using GWR model with kernel
adaptive bi-square weighting and comparing it to the conventional linear
regression model. The data used in the
study are secondary data obtained from the National Socio-Economic Survey and
Basic Health Research (2010) conducted in 38 districts in East Java. Estimation
is done by using the Weighted Least Squares method that provides different
weighting values to each region. The result showed that there are 38 models of
the malnutrition case that is different for each district in East Java. The GWR
model with bi-square kernel weighting function is better in modelling the case
of malnutrition in East Java compared to the conventional linear regression
models that are based on the criteria of goodness that is the R-square, Mean
Square Error and the Akaike Information Criterion.
Keywords: geographical
weighted regression; malnutrition; weighted least square
Author: Ida Mariati Hutabarat,
Asep Saefuddin, Hardinsyah, Anik Djuraidah
Journal Code: jpkedokterangg150381
