Electronic Theses and Dissertation

Penelitian ini bertujuan untuk menentukan model terbaik antara regresi data panel dan Geographically Weighted Panel Regression (GWPR) dalam menjelaskan persentase penduduk miskin di Provinsi Aceh serta mengidentifikasi faktor-faktor yang signifikan memengaruhi persentase penduduk miskin di daerah tersebut. GWPR merupakan modifikasi dari model regresi yang menggabungkan GWR dan regresi data panel. Regresi data panel digunakan sebagai pendekatan awal untuk mengevaluasi pengaruh global variabel-variabel terhadap persentase penduduk miskin, dengan mempertimbangkan data yang bersifat cross-section berupa data 23 kabupaten/kota dan time series dari tahun 2018 - 2022. Metode GWPR kemudian digunakan untuk menganalisis variasi pengaruh variabel-variabel secara spasial dan temporal di setiap wilayah. Hasil analisis menunjukkan bahwa model GWPR lebih unggul dibandingkan model regresi data panel REM, dengan nilai R² terbesar sebesar 94,29%, nilai AIC terkecil sebesar 41,830, dan nilai RSS terkecil sebesar 6,499. Selain itu, analisis GWPR mengidentifikasi faktor-faktor yang secara spasial signifikan memengaruhi persentase penduduk miskin di Provinsi Aceh. 5 variabel yang paling sering signifikan adalah angka harapan hidup (X3) dan tingkat pengangguran terbuka (X4), yang signifikan di 15 wilayah, rata-rata lama sekolah (X1) di 13 wilayah, serta harapan lama sekolah (X2) dan jumlah penduduk (X8) di 11 wilayah. Kabupaten Aceh Besar, Aceh Jaya, Kota Banda Aceh, dan Pidie memiliki jumlah variabel signifikan terbanyak. Secara keseluruhan, terdapat 17 kombinasi variabel signifikan di berbagai wilayah.

Kata kunci: Geographically Weighted Panel Regression, Regresi Data Panel, Persentase Penduduk Miskin, Faktor-faktor Spasial.

This study aims to determine the best model between panel data regression and Geographically Weighted Panel Regression (GWPR) in explaining the percentage of the poor population in Aceh Province and to identify the significant factors affecting the percentage of the poor population in the region. GWPR is a modification of the regression model that combines GWR and panel data regression. Panel data regression is used as an initial approach to evaluate the overall influence of the variables on the percentage of the poor population, considering the data in the form of cross-section from 23 districts/cities and time series data from 2018 - 2022. The GWPR method is then applied to analyze the varying effects of these variables spatially and temporally in each area. The analysis results indicate that the GWPR model outperforms the REM panel data regression model, with the highest R² value of 94,29%, the lowest AIC value of 41,830, and the lowest RSS value of 6,499. Additionally, the GWPR analysis identified factors that are spatially significant in influencing the percentage of the poor population in Aceh Province. The five most frequently significant variables are life expectancy (X3) and the open unemployment rate (X4), which are significant in 15 regions, average years of schooling (X1) in 13 regions, as well as expected years of schooling (X2) and population (X8) in 11 regions. Aceh Besar, Aceh Jaya, Banda Aceh City, and Pidie have the highest number of significant variables. Overall, there are 17 combinations of significant variables across different regions. Keywords: Geographically Weighted Panel Regression, Panel Data Regression, Percentage of Poor Population, Spatial Factors.