Pemodelan Jumlah Kematian Akibat Demam Berdarah Dengue di Provinsi Jawa Barat dan Banten Menggunakan Geographically Weighted Generalized Poisson Regression

Rosyida, Rahmalia (2026) Pemodelan Jumlah Kematian Akibat Demam Berdarah Dengue di Provinsi Jawa Barat dan Banten Menggunakan Geographically Weighted Generalized Poisson Regression. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003221112-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Demam Berdarah Dengue (DBD) masih menjadi salah satu masalah kesehatan masyarakat di Indonesia dengan jumlah kasus dan kematian yang bervariasi antarwilayah. Perbedaan karakteristik geografis, lingkungan, dan fasilitas kesehatan menyebabkan faktor-faktor yang memengaruhi jumlah kematian akibat DBD tidak selalu sama pada setiap daerah. Penelitian ini bertujuan memodelkan jumlah kematian akibat DBD pada kabupaten/kota di Provinsi Jawa Barat dan Banten tahun 2024 menggunakan metode Geographically Weighted Generalized Poisson Regression (GWGPR). Variabel respon yang digunakan adalah jumlah kematian akibat DBD, sedangkan variabel prediktor meliputi kepadatan penduduk, rata-rata ketinggian wilayah, total curah hujan, persentase rumah tangga dengan akses sanitasi aman, persentase rumah tangga dengan akses air minum layak, dan rasio fasilitas kesehatan. Analisis diawali dengan pemodelan Generalized Poisson Regression (GPR) untuk mengatasi overdispersi, kemudian dilanjutkan dengan pemodelan GWGPR menggunakan pembobot Adaptive Bisquare Kernel dan Adaptive Gaussian Kernel. Pemilihan model terbaik dilakukan berdasarkan nilai Akaike Information Criterion corrected (AICc). Hasil penelitian menunjukkan bahwa data jumlah kematian akibat DBD mengalami overdispersi dengan parameter dispersi sebesar 5,19 serta terdapat heterogenitas spasial berdasarkan uji Breusch-Pagan. Pada model GPR, variabel kepadatan penduduk, rata-rata ketinggian wilayah, total curah hujan, persentase rumah tangga dengan akses air minum layak, dan rasio fasilitas kesehatan berpengaruh signifikan terhadap jumlah kematian akibat DBD. Berdasarkan kriteria AICc, model GWGPR dengan pembobot Adaptive Gaussian Kernel merupakan model terbaik dengan nilai AICc, lebih kecil dibandingkan model GWGPR Adaptive Bisquare Kernel dan model GPR. Model terbaik menghasilkan dua kelompok wilayah berdasarkan kombinasi variabel signifikan, yang menunjukkan adanya variasi spasial faktor-faktor yang memengaruhi jumlah kematian akibat DBD. Dengan demikian, pendekatan GWGPR Adaptive Gaussian Kernel mampu memberikan pemodelan yang lebih baik dibandingkan model global dalam menangkap keragaman karakteristik antarwilayah.
============================================================
====================================================================
Dengue Hemorrhagic Fever (DHF) remains a public health problem in Indonesia, with varying numbers of cases and deaths across regions. Differences in geographic characteristics, environment, and health facilities mean that factors influencing the number of DHF deaths vary across regions. This study aims to model the number of DHF deaths in districts/cities in West Java and Banten Provinces in 2024 using the Geographically Weighted Generalized Poisson Regression (GWGPR) method. The response variable is the number of DHF deaths, while the predictor variables include population density, average regional altitude, rainfall, percentage of households with access to safe sanitation, percentage of households with access to improved drinking water, and the ratio of health facilities. The analysis begins with Generalized Poisson Regression (GPR) modeling to address overdispersion, then continues with GWGPR modeling using Adaptive Bisquare Kernel and Adaptive Gaussian Kernel weighting. The best model is selected based on the Akaike Information Criterion corrected (AICc) value. The results showed that the data on the number of dengue fever deaths experienced overdispersion with a dispersion parameter of 5.19 and there was spatial heterogeneity based on the Breusch-Pagan test. In the GPR model, the variables of population density, average altitude of the area, amount of rainfall, percentage of households with access to clean drinking water, and the ratio of health facilities had a significant effect on the number of dengue fever deaths. Based on the AICc criteria, the GWGPR model with Adaptive Gaussian Kernel weighting was the best model with an AICc, smaller than the GWGPR Adaptive Bisquare Kernel model and the GPR model. The best model produced two groups of regions based on a combination of significant variables, indicating spatial variation in factors that influence the number of dengue fever deaths. Thus, the GWGPR Adaptive Gaussian Kernel approach is able to provide better modeling than the global model in capturing the diversity of characteristics between regions.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	DFH, GWGPR, West Java and Banten, Spatial, DFH, GWGPR, West Java and Banten, Spatial
Subjects:	H Social Sciences > HA Statistics > HA30.6 Spatial analysis Q Science > Q Science (General) Q Science > QA Mathematics
Divisions:	Faculty of Mathematics and Science > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Rahmalia Rosyida
Date Deposited:	29 Jul 2026 07:55
Last Modified:	29 Jul 2026 07:55
URI:	http://repository.its.ac.id/id/eprint/138940

Actions (login required)

View Item