Work

Analysis of Geographically Weighted Logistic Regression Models with A Bisquare Weighting Matrix on Poverty Status in West Java

Year: 2025

Type: article

Abstract: This research addresses the first Sustainable Development Goal and aims to analyze poverty status in West Java Province, which has the second highest number of poor people in Indonesia. The study employs Geographically Weighted Logistic Regression (GWLR) and compares it with global logistic regression. Influential variables include GDP, unemployment, population density, access to safe water, and roof type (bamboo/wood). Results show that 55.6% of regions are classified as poor, with the GWLR model using a Fixed Bisquare kernel achieving 81.4% accuracy, outperforming global logistic regression at 66.7%. Significant variables vary by region: unemployment rate in Bogor, Depok, and Bekasi; population density in Bekasi, Karawang, and Purwakarta; water access in Sukabumi; and roof type in Indramayu and Bogor. These spatial variations suggest that poverty reduction requires a region-specific approach. Consequently, policies should be formulated considering the priorities and characteristics of each region in West Java Province.

Source: CAUCHY Jurnal Matematika Murni dan Aplikasi

DOI: 10.18860/cauchy.v10i2.36315