PEMODELAN INDEKS PEMBANGUNAN MANUSIA DI PROVINSI NUSA TENGGARA TIMUR MENGGUNAKAN GEOGRAPHICALLY WEIGHTED REGRESSION

Abstract

The human development index of East Nusa Tenggara Province still ranks third lowest at 65.28%. Differences in regional locations in East Nusa Tenggara are suspected to have a spatial effect on the HDI. This suspected spatial diversity in achieving the HDI presents challenges; specifically, observational data may struggle to meet the classical regression assumption of residual variance homogeneity. To address this issue, a spatial approach is used to model the HDI in the region. In this context, Geographically Weighted Regression (GWR) is applied as it takes into account aspects of geographic location. The purpose of this study is to model the HDI using the adaptive bisquare kernel weighting function and to determine the best model in East Nusa Tenggara Province. The data used are drawn from the official website of the Central Statistics Agency. Upon testing, spatial heterogeneity was identified in the HDI across 21 districts and 1 city in East Nusa Tenggara Province. As a result, it is appropriate to continue modeling with the GWR approach. Based on the analysis results, 22 different models were obtained for each region, and the best GWR model was found in Rote Ndao Regency with a coefficient of determination of 99.49%. This means that 99.49% of the variation in the human development index in the region is explained by the study's variables, while the remaining 0.51% is attributed to factors outside the research variables.

Keywords: Adaptive bisquare kernel; Geographically Weighted Regression; Human Development Index