Open AccessDistance weight of GWR-Kriging model for stunting cases in East Java
Author(s) -
Deby Ardianti,
Henny Pramoedyo,
N Nurjannah
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1968/1/012028
Subject(s) - kriging , variogram , statistics , weighting , geographically weighted regression , kernel (algebra) , closeness , regression analysis , gaussian function , ordinary least squares , gaussian , mathematics , medicine , mathematical analysis , physics , combinatorics , quantum mechanics , radiology
The chosen of distance weights is needed to form an accurate Geographically Weighting Regression model. There are 3 types of distance weights namely Gaussian kernel, Bisquare kernel and Tricube kernel. The weighting in GWR describes the closeness relation between locations. For data that has spatial heterogeneity, GWR models are more suitable models than OLS models. This study was conducted to obtain the best distance weighting based on minimum cross-validation method. Using secondary data from the Health Department in East Java with 34 districts for observation, the dependent variable is stunting and five independent variables that influence stunting cases. Based on the result, GWR models with fixed gaussian models produces a better accuracy in higher R 2 values compared to OLS models. The predicted map of the spread stunting cases conducted by interpolation GWR Kriging using exponential semivariogram.