Chotimah, Chusnul (2019) Pemodelan Ketimpangan Pendapatan di Jawa Timur Menggunakan Geographically Weighted Panel Regression. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
06211750017006-Master_Thesis.pdf - Accepted Version Download (6MB) | Preview |
Abstract
Analisis regresi merupakan salah satu metode statistik yang mempelajari hubungan antara variabel respon dengan variabel prediktor. Estimasi parameter pada regresi linier klasik menghasilkan koefisien regresi yang diduga berlaku global untuk keseluruhan unit observasi. Beberapa kasus, kondisi antar lokasi satu dan lainnya berbeda karena pengaruh aspek spasial yang menyebabkan heterogenitas spasial. Geographically Weighted Regression (GWR) adalah teknik regresi lokal pengembangan dari regresi klasik dengan menggunakan data spasial. Dalam suatu penelitian juga dibutuhkan data yang melibatkan data cross section dan time series atau disebut sebagai data panel. Geographically Weighted Panel Regression (GWPR) merupakan gabungan antara GWR dan regresi data panel. Model GWPR ini selanjutnya digunakan untuk pemodelan ketimpangan pendapatan di Provinsi Jawa Timur pada tahun 2010-2014. Hasil penelitian menunjukkan bahwa model GWPR dengan menggunakan pembobot kernel adaptive gaussian menghasilkan goodness of fit yang lebih baik dari model FEM within estimator, dan model yang dihasilkan setiap lokasi berbeda antara satu sama lainnya. Variabel yang signifikan mempengaruhi ketimpangan pendapatan di Jawa Timur yaitu persentase PDRB pertanian, kehutanan, dan perikanan, persentase PDRB industri pengolahan, persentase PDRB PMTB, persentase PDRB perdagangan besar dan eceran, reparasi dan perawatan mobil dan sepeda motor, persentase PDRB informasi dan komunikasi. Model GWPR menghasilkan nilai R2 sebesar 91,02%, dengan Mean Square Error (MSE) sebesar 0,0004.
======================================================================================================================================
Regression analysis is a common statistical method which widely used to evaluate the relationship between variables. Parameter estimates in classical linear regression produce regression coefficients which imply the effect of the predictor to respond to the entire observation unit. In some cases, the conditions between one and another location are different because of the influence of spatial aspects that cause spatial heterogeneity. Geographically Weighted Regression (GWR) is the development of classical regression approach for spatial data. This type of study needs data involving cross-section and time-series data or referred to as panel data. Geographically Weighted Panel Regression (GWPR) is a combination of GWR and panel data regression. In this study, GWPR model is used to modeling income inequality in East Java Province in 2010-2014. The results showed that the GWPR model using an adaptive gaussian kernel weighted resulted in better goodness of fit than the FEM model within the estimator, and the models produced by each location differed from one another. Significant variables affect income inequality in East Java, namely the percentage of GDP Regional of agriculture, forestry and fisheries, the percentage of GDP Regional processing industry, the percentage of GDP Regional of gross fixed capital formation, the percentage of GDP Regional of large and retail trade, repair and maintenance of cars and motorbikes, the percentage of GDP Regional of information and communication. The GWPR model produces R2 value of 91.02%, with a Mean Square Error (MSE) of 0.0004.
| Item Type: | Thesis (Masters) |
|---|---|
| Additional Information: | RTSt 519.536 Cho p-1 2019 |
| Uncontrolled Keywords: | GWR, GWPR, Ketimpangan Pendapatan, Regresi Data Panel |
| Subjects: | H Social Sciences > HA Statistics > HA30.6 Spatial analysis |
| Divisions: | Faculty of Mathematics and Science > Statistics > 49101-(S2) Master Thesis |
| Depositing User: | CHOTIMAH CHUSNUL |
| Date Deposited: | 15 Dec 2023 07:26 |
| Last Modified: | 15 Dec 2023 07:26 |
| URI: | http://repository.its.ac.id/id/eprint/68635 |
Actions (login required)
 |
View Item |