PERBANDINGAN ADAPTIVE GAUSSIAN DAN FIXED GAUSSIAN PADA GEOGRAPHICALLY WEIGHTED LOGISTIC REGRESSION DALAM PEMODELAN KEMISKINAN DI JAWA BARAT
Kode Repository :SKM9/HAN/23
NPM :064119012
Nama :Hanna Utari Ayunintias
Pembimbing 1 :-Hagai Wijayanti, S.Si., M.Si
Pembimbing 2 :-Isti Kamila, S.Pd., M.Si
Abstrak :-Abstract: West Java Province is a province faced with the problem of poverty. The
percentage of poverty in West Java in 2022 increased by 0.01% from 2021. Poverty in
West Java involves regional effects, so the appropriate method to analyze poverty in West
Java is Geographically Weighted Logistic Regression (GWLR). This study aims to
compare the adaptive gaussian kernel and fixed gaussian kernel weighting functions in
the GWLR model in modeling poverty cases in West Java in 2022. In the best model,
significant factors affecting poverty in West Java can be identified. This study uses data
on the percentage of poor people (Y) and the factors that influence it, namely population
(
), minimum wage (
), human development index (
), PDRB per capita (
),
and open unemployment rate (
) in 27 cities/regencies in West Java. The results
showed that the GWLR model with fixed gaussian kernel weighting function was the best
model in modeling poverty in West Java in 2022 based on the smallest Akaike
Information Criterion (AIC) value of 33.824. The significant factor affecting poverty for
the fixed gaussian kernel weighting function is the human development index in the
observation locations of Bogor Regency, Sukabumi Regency, Cianjur Regency,
Purwakarta Regency, Karawang Regency, Bekasi Regency, West Bandung Regency,
Bogor City, Sukabumi City, Bandung City, Bekasi City, and Depok City.
Keywords: poverty, GWLR, fixed gaussian kernel, adaptive gaussian kerne