PEMODELAN MULTISCALE GEOGRAPHICALLY WEIGHTED REGRESSION (MGWR) UNTUK ANALISIS ANGKA BUTA HURUF DI PROVINSI SUMATERA SELATAN TAHUN 2021

• Main Author: Rizky Ardhani, - (Author)
• Format: Book
• Published: 2023-04-17.
• Subjects: Thesis; NonPeerReviewed
• Online Access: Link Metadata
• Notes: http://repository.upi.edu/89820/1/S_MAT_1900507_Title.pdf; http://repository.upi.edu/89820/2/S_MAT_1900507_Chapter1.pdf; http://repository.upi.edu/89820/3/S_MAT_1900507_Chapter2.pdf; http://repository.upi.edu/89820/4/S_MAT_1900507_Chapter3.pdf; http://repository.upi.edu/89820/5/S_MAT_1900507_Chapter4.pdf; http://repository.upi.edu/89820/6/S_MAT_1900507_Chapter5.pdf; http://repository.upi.edu/89820/7/S_MAT_1900507_Appendix.pdf
• Keywords: L Education (General); QA Mathematics

Multiscale Geographically Weighted Regression (MGWR) merupakan suatu metode regresi dengan memperhatikan aspek spasial, yaitu heterogenitas spasial yang dipengaruhi oleh kondisi geografis dari masing-masing lokasi pengamatan. Metode MGWR dikembangkan dari Geographically Weighted Regression (GWR), karena permasalahan yang sering dihadapi seperti penggunaan satu bandwidth dan ketidakmampuan mengatasi multikolinearitas. Bandwidth MGWR memungkinkan untuk digunakan pada tiap variabel, sehingga diharapkan mampu memberikan suatu ketepatan model yang diduga lebih akurat terhadap suatu data. Pada penelitian ini, metode MGWR diterapkan pada fenomena Angka Buta Huruf (ABH) yang masih terjadi di Provinsi Sumatera Selatan. Berdasarkan hasil analisis, diperoleh 17 model dari 17 lokasi pengamatan dengan nilai R^2 sebesar 0,862. Salah satu model ABH yang terdapat di Kota Palembang diperoleh nilai koefisien intercept sebesar -0,0187, koefisien β_1 sebesar 0,558, koefisien β_2 sebesar -0,1748, koefisien β_3 sebesar -0,0062, koefisien β_4 sebesar -1,6129, koefisien β_5 sebesar -1,4489, dan koefisien β_6 sebesar 0,5394. Multiscale Geographically Weighted Regression (MGWR) is a regression method by taking into account the spatial aspects, that are influenced by the geographical conditions of each observation location. The MGWR method was developed from Geographically Weighted Regression (GWR), because of problems that are often encountered such as the use of one bandwidth and the inability to overcome multicollinearity. The MGWR bandwidth allows it to be used for each variable, so that it is expected to be able to provide a model accuracy that is thought to be more accurate for a data. In this study, the MGWR method was applied to the phenomenon of Illiteracy Rate (ABH) which still occurs in South Sumatra Province. Based on the results of the analysis, 17 models were obtained from 17 observation locations with an R^2 value of 0.862. One of the ABH models found in Palembang City obtained an intercept coefficient of -0.0187, β_1 coefficient of 0.558, β_2 coefficient of -0.1748, β_3 coefficient of -0.0062, β_4 coefficient of -1.6129, β_5 coefficient of -1.4489, and the coefficient β_6 of 0.5394.