Model Geographically Weighted Bivariate Poisson Inverse Gamma Regression (Studi Kasus: Jumlah Kematian Ibu dan Kematian Bayi Neonatal di Provinsi Jawa Timur)

Madani, Ahmad (2025) Model Geographically Weighted Bivariate Poisson Inverse Gamma Regression (Studi Kasus: Jumlah Kematian Ibu dan Kematian Bayi Neonatal di Provinsi Jawa Timur). Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6003222007-Masther_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (2MB) | Request a copy

Abstract

Tujuan penelitian ini adalah mengembangkan model regresi mixed Poisson untuk menangani kasus overdispersi pada data cacahan, yaitu model Bivariate Poisson Inverse Gamma Regression (BPIGAR) yang melibatkan variabel eksposur. Selanjutnya, model BPIGAR dikembangkan menjadi model Geographically Weighted Bivariate Poisson Inverse Gamma Regression (GWBPIGAR) dengan memasukkan efek lokasi pada model. Kajian teori dilakukan untuk mendapatkan penaksir parameter model BPIGAR dan GWBPIGAR menggunakan metode Maximum Likelihood Estimation (MLE) dengan iterasi algoritma Bern, Hall, Hall, dan Hausman (BHHH). Selanjutnya, model BPIGAR dan GWBPIGAR diaplikasikan untuk memodelkan jumlah kasus kematian ibu dan bayi neonatal di Provinsi Jawa Timur pada tahun 2022 menggunakan lima variabel prediktor dan dua variabel eksposure, serta membandingkan dua pembobot yaitu kernel Adaptive Gaussian dan kernel Adaptive Bisquare. Hasil penelitian menunjukkan bahwa pada model BPIGAR, variabel prediktor yang berpengaruh signifikan terhadap kematian ibu adalah persentase ibu hamil yang mendapat tablet tambah darah, persentase persalinan yang ditolong oleh tenaga keehatan dan persentase komplikasi kebidanan yang ditangani, sedangkan untuk jumlah kematian bayi neonatal semua variabel prediktor berpengaruh signifikan terhadap jumlah kematian bayi neonatal. Selanjutnya pada model GWBPIGAR, jumlah kematian ibu membagi kabupaten/kota menjadi 9 kelompok dan untuk jumlah kematian bayi neonatal membagi kelompok kabupaten/kota menjadi 3 kelompok untuk pembobot kernel Adaptive Gaussian. Berdasarkan nilai AICc, model GWBPIGAR dengan pembobot kernel Adaptive Gaussian lebih baik dari pembobot Adaptive Bisquare dalam memodelkan data jumlah kematian ibu dan kematian bayi neonatal di Provinsi Jawa Timur pada Tahun 2022.
====================================================================================================================================
The purpose of this study is to develop a mixed Poisson regression model to handle overdispersion cases in count data, namely the Bivariate Poisson Inverse Gamma Regression (BPIGAR) model involving exposure variables. Furthermore, the BPIGAR model was developed into a Geographically Weighted Bivariate Poisson Inverse Gamma Regression (GWBPIGAR) model by including location effects in the model. Theoretical studies were conducted to obtain parameter estimators of the BPIGAR and GWBPIGAR models using the Maximum Likelihood Estimation (MLE) method with the Bern, Hall, Hall, and Hausman (BHHH) iteration algorithm. Furthermore, the BPIGAR and GWBPIGAR models were applied to model the number of cases of maternal and neonatal deaths in East Java Province in 2022 using five predictor variables and two exposure variables, and comparing two weights namely Adaptive Gaussian kernel and Adaptive Bisquare kernel. The results showed that in the BPIGAR model, the predictor variables that had a significant effect on maternal mortality were the percentage of pregnant women who received blood-added tablets, the percentage of deliveries assisted by health workers and the percentage of obstetric complications treated, while for the number of neonatal deaths all predictor variables had a significant effect on the number of neonatal deaths. Furthermore, in the GWBPIGAR model, the number of maternal deaths divides districts/cities into 9 groups and for the number of neonatal infant deaths divides districts/cities into 3 groups for Adaptive Gaussian kernel weights. Based on the AICc value, the GWBPIGAR model with Adaptive Gaussian kernel weights is better than Adaptive Bisquare weights in modeling data on the number of maternal deaths and neonatal infant deaths in East Java Province in 2022.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	BPIGAR, BHHH, Eksposur, GWBPIGAR, MLE dan MLRT
Subjects:	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:	Ahmad Madani
Date Deposited:	10 Jan 2025 06:25
Last Modified:	10 Jan 2025 06:25
URI:	http://repository.its.ac.id/id/eprint/116259

Actions (login required)

View Item