IMPLEMENTATION OF MIXED GEOGRAPHICALLY WEIGHTED REGRESSION MODEL TO ANALYZE SOCIAL ASSISTANCE BUDGET IN EAST JAVA

Abstract

Background - Social assistance (BANSOS) is aid provided by the government to low-income communities in the form of money, goods, or services. Understanding the allocation and influencing factors of social assistance in East Java is crucial for effective distribution. Mixed Geographically Weighted Regression (MGWR) combines global and local regression models to address spatial variability in the data.

Purpose – This study aims to develop an MGWR model with a fixed kernel weighting function for the social assistance budget in East Java for 2022. The specific objectives are to identify factors affecting the budget and determine the best model that represents these global and local relationships.

Methodology – The study employs the Mixed Geographically Weighted Regression (MGWR) method with a fixed Gaussian kernel to analyze social assistance budget data and economic factors in East Java for 2022. Models OLS, GWR, and MGWR are applied and evaluated using the Akaike Information Criterion (AIC) to identify the best-performing model.

Findings – The MGWR model with a fixed Gaussian kernel is the best for the social assistance budget in East Java, yielding a lower AIC compared to OLS and GWR models. The globally influential factor in this model is economic growth (