Model Simultan Dependensi Dan Heterogenitas Spasial Dengan MGWR-SAR (Studi Kasus: Pemodelan Harga Rumah di DKI Jakarta)

Sagala, Yenro P (2026) Model Simultan Dependensi Dan Heterogenitas Spasial Dengan MGWR-SAR (Studi Kasus: Pemodelan Harga Rumah di DKI Jakarta). Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6003232003_Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (11MB) | Request a copy

Abstract

Pemodelan spasial menjadi pendekatan esensial dalam analisis data dengan dependensi dan heterogenitas spasial. Aspek tersebut dapat dideteksi dengan uji Moran’s I dan Uji Breusch-Pagan, dan umum dimodelkan dengan Spatial Autoregressive (SAR) serta Geographically Weighted Regression (GWR). Dalam beberapa kasus, kedua aspek ini muncul secara bersamaan, tetapi pendekatan metodologi yang secara simultan memodelkan heterogenitas dan dependensi spasial masih terbatas. Untuk mengatasi keterbatasan tersebut, Mixed Geographically Weighted Regression-Spatial Autoregressive (MGWR-SAR), model integrasi fleksibilitas parameter lokal dari MGWR dengan struktur dependensi spasial dari SAR. Hasil menunjukkan bahwa metode Spatial Two-Stage Least Square (S2SLS) dapat diimplementasikan untuk estimasi parameter model MGWR-SAR. Di antara model yang dievaluasi, model Mixed Geographically Weighted Regression- Global Spatial Autoregressive (MGWR-GSAR) Threshold yang model MGWR-SAR dengan spesifikasi kernel adaptive bisquare dan bobot spasial berbasis threshold distance yang diasumsikan bersifat global, terbukti lebih unggul dari pada model lainnya dalam memodelkan keragaman harga rumah di DKI Jakarta. Pendekatan ini secara efektif menangkap keragaman pengaruh karakteristik rumah dan karakteristik lokasi antarwilayah serta pengaruh harga rumah di lokasi yang bertetangga terhadap harga rumah. Seluruh variabel karakteristik rumah menunjukkan hubungan kausal positif dengan harga rumah, dan hampir seluruh karakteristik lokasi menunjukkan hubungan kausal negatif seperti yang diharapkan kecuali jarak ke fasilitas kesehatan dan pusat transportasi umum. Hasil tersebut menunjukkan bahwa MGWR-GSAR Threshold menjadi pendekatan yang robust untuk memodelkan secara simultan dependensi dan heterogenitas spasial pada fenomena kompleks seperti harga rumah di DKI Jakarta.
===================================================================================================================================
Spatial modeling is a critical approach for analyzing data characterized by spatial dependence and heterogeneity. This data can be identified using Moran's I and Breusch–Pagan tests and commonly modeled through Spatial Autoregressive (SAR) and Geographically Weighted Regression (GWR) frameworks. In many applications, these two spatial aspects occur simultaneously; however, methodological approaches capable of addressing both at once remain limited. To address this challenge, the Mixed Geographically Weighted Regression–Spatial Autoregressive (MGWR-SAR) model integrates the local parameter flexibility of MGWR with the spatial dependence structure of SAR. The findings indicate that the Spatial Two-Stage Least Squares (S2SLS) method can be effectively applied to estimate the parameters of the MGWR-SAR model. Among the models evaluated, the model Mixed Geographically Weighted Regression- Global Spatial Autoregressive (MGWR-GSAR) Threshold, a MGWR-SAR specification employing an adaptive bisquare kernel and a globally assumed threshold-distance spatial weight matrix, demonstrates superior explanatory performance on explaing the vaiablitiy of house price in DKI Jakarta. This approach effectively captures the spatially varying influences of housing and locational characteristics, as well as the spillover effects of neighboring housing prices. All housing characteristic variables exhibit a positive causal relationship with house prices, and nearly all location characteristics show a negative causal relationship as expected, except for the distance to healthcare facilities and public transportation hubs. The results demonstrate that the MGWR-GSAR Threshold model offers a robust framework for simultaneously addressing spatial dependence and heterogeneity in complex urban phenomena, such as housing prices in DKI Jakarta.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Dependensi Spasial, Harga Rumah, Heterogenitas Spasial, MGWR-SAR, S2SLS
Subjects:	H Social Sciences > HA Statistics > HA30.6 Spatial analysis
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis
Depositing User:	Yenro P. Sagala
Date Deposited:	19 Jan 2026 08:57
Last Modified:	19 Jan 2026 08:57
URI:	http://repository.its.ac.id/id/eprint/129756

Actions (login required)

View Item