Nurdin, Sitti Zahirah (2025) Model Regresi Geographically Weighted Generalized Log Gamma (Studi Kasus: Angka Kematian Bayi di Provinsi Sulawesi Tenggara, Sulawesi Tengah dan Gorontalo Tahun 2022). Masters thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 6003222014-Master_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
6003222014-Master_Thesis.pdf Restricted to Repository staff only until 1 April 2027. Download (5MB) | Request a copy |
Abstract
Model Generalized Log Gamma Regression (GLGR) adalah model regresi yang digunakan ketika variabel respon mengikuti distribusi generalized log gamma. Model ini bersifat global, di mana penaksir parameter yang dihasilkan sama untuk semua unit pengamatan. Namun, pada data berbasis wilayah, efek spasial berupa heterogenitas spasial sering kali terjadi, sehingga model global menjadi kurang tepat. Penelitian ini mengembangkan model GLGR yang melibatkan faktor spasial dengan pembobot geografis, yang dikenal sebagai model Geographically Weighted Generalized Log Gamma Regression (GWGLGR). Penaksiran parameter dalam model dilakukan menggunakan Maximum Likelihood Estimation (MLE) yang diikuti iterasi numerik dengan metode Berndt-Hall-Hall-Hausman (BHHH). Pengujian hipotesis dilakukan secara serentak menggunakan Maximum Likelihood Ratio Test (MLRT) dan secara parsial menggunakan uji Z. Model GWGLGR diterapkan pada data Angka Kematian Bayi (AKB) di Provinsi Sulawesi Tenggara, Sulawesi Tengah, dan Gorontalo tahun 2022. Hasil penelitian menunjukkan bahwa model GWGLGR memiliki nilai AICc sebesar 144,4299 yang lebih rendah dibandingkan model GLGR yaitu sebesar 254,2492 yang menandakan bahwa model GWGLGR lebih baik dalam memodelkan angka kematian bayi. Selain itu, pemodelan GWGLGR menghasilkan enam kelompok kabupaten/kota berdasarkan variabel prediktor yang signifikan terhadap variabel respon.
===================================================================================================================================
The Generalised Log Gamma Regression (GLGR) model is a regression model used when the response variable follows a generalised log gamma distribution. This model is global, where the resulting parameter estimator is the same for all observation units. However, in region-based data, spatial effects in the form of spatial heterogeneity often occur, making the global model less appropriate. This research develops a GLGR model that involves spatial factors with geographical weighting, known as the Geographically Weighted Generalised Log Gamma Regression (GWGLGR) model. Parameter estimation in the model is done using Maximum Likelihood Estimation (MLE) followed by numerical iteration with the Berndt-Hall-Hausman (BHHH) method. Hypothesis testing was conducted simultaneously using the Maximum Likelihood Ratio Test (MLRT) and partially using the Z test. The GWGLGR model was applied to Infant Mortality Rate (IMR) data in Southeast Sulawesi, Central Sulawesi, and Gorontalo provinces in 2022. The results showed that the GWGLGR model has an AICc value of 144,4299 which is lower than the GLGR model of 245.2492, indicating that the GWGLGR model is better at modelling infant mortality. In addition, GWGLGR modelling produced six groups of districts/cities based on predictor variables that were significant to the response variable.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | AKB, BHHH, GLGR, GWGLGR, MLRT. BHHH, GLGR, GWGLGR, IMR, MLRT. |
| Subjects: | H Social Sciences > HA Statistics > HA29 Theory and method of social science statistics H Social Sciences > HA Statistics > HA30.6 Spatial analysis H Social Sciences > HA Statistics > HA31.3 Regression. Correlation. Logistic regression analysis. H Social Sciences > HA Statistics > HA31.7 Estimation |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis |
| Depositing User: | Sitti Zahirah Nurdin |
| Date Deposited: | 07 Feb 2025 03:30 |
| Last Modified: | 07 Feb 2025 03:30 |
| URI: | http://repository.its.ac.id/id/eprint/118529 |
Actions (login required)
 |
View Item |