Muchlisun, Alfaro Alamsyah (2026) Implementasi Geographically Weighted Generalized Poisson Regression dengan Optimasi Fisher-Scoring untuk Pemodelan Kasus Gizi Buruk Balita di Jawa Tengah. Undergraduate thesis, UPN Veteran Jawa Timur.
|
Text (Cover)
22083010041-cover.pdf Download (1MB) | Preview |
|
|
Text (Bab 1)
22083010041-bab1.pdf Download (221kB) | Preview |
|
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Bab 2)
22083010041-bab2.pdf Restricted to Repository staff only until 19 May 2028. Download (570kB) |
|
|
Text (Bab 3)
22083010041-bab3.pdf Restricted to Repository staff only until 19 May 2028. Download (524kB) |
|
|
Text (Bab 4)
22083010041-bab4.pdf Restricted to Repository staff only until 19 May 2028. Download (1MB) |
|
|
Text (Bab 5)
22083010041-bab5.pdf Download (189kB) | Preview |
|
|
Text (Daftar Pustaka)
22083010041-daftarpustaka.pdf Download (156kB) | Preview |
|
|
Text (Lampiran)
22083010041-lampiran.pdf Restricted to Repository staff only until 19 May 2028. Download (55kB) |
Abstract
Cases of malnutrition among toddlers in Central Java Province remain a serious health issue. In 2022, approximately 9,499 infants (1% of the total infants weighed) were recorded as suffering from malnutrition, the highest rate on the island of Java. Although the percentage dropped to 0.64% in 2024, the number of cases actually increased to 11,348 infants, indicating that the number of malnutrition cases remains high and has not yet shown a decline. From a data perspective, the response variable consists of count data, which is typically modeled using Poisson regression. However, the analysis results indicate that the variance is significantly greater than the mean, with a dispersion ratio of 162.5811 (>1), indicating overdispersion and rendering the Poisson model less appropriate. Additionally, the Breusch-Pagan test indicates the presence of spatial heterogeneity (BP = 8.1539; p-value = 0.04294), meaning the influence of risk factors varies across regions. To address this, the Geographically Weighted Generalized Poisson Regression (GWGPR) method was used, which is capable of handling overdispersion while accounting for differences in effects across regions. This model works by assigning weights based on geographical proximity, so that parameters are estimated locally. The estimation process is performed using the Fisher-Scoring algorithm, which iteratively updates the parameters until convergence. The results show that GWGPR performs best (AIC 496.374; BIC 497.133), outperforming GPR (AIC 505.678; BIC 513.455) and Poisson Regression (AIC 5716.718; BIC 5722.939). The variable for the percentage of active posyandu (X2) was significant across all regions, while iron tablet consumption (X4) and poverty (X5) were locally significant. Meanwhile, the variable for complete basic immunization (X1), as a categorical variable, did not show a significant effect in the model. Overall, GWGPR provided more specific results in explaining the risk factors for malnutrition.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | H Social Sciences > HA Statistics | ||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Data Science | ||||||||||||
| Depositing User: | Alfaro Alamsyah Muchlisun | ||||||||||||
| Date Deposited: | 19 May 2026 06:48 | ||||||||||||
| Last Modified: | 19 May 2026 06:48 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/51636 |
Actions (login required)
 |
View Item |