Pemodelan Multivariate Adaptive Bivariate Generalized Poisson Regression Spline (MABGPRS) (Studi Kasus: Stunting Dan Wasting Di Sulawesi Tenggara)

Sadhana, Tharatya Kartika (2024) Pemodelan Multivariate Adaptive Bivariate Generalized Poisson Regression Spline (MABGPRS) (Studi Kasus: Stunting Dan Wasting Di Sulawesi Tenggara). Masters thesis, INSTITUT TEKNOLOGI SEPULUH NOPEMBER.

Text 6003221027_Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (2MB) | Request a copy

Abstract

Multivariate Adaptive Regression Spline (MARS) merupakan regresi non-parametrik yang mampu mengatasi data dengan banyak variabel prediktor. Metode MARS dapat diterapkan pada data dengan pola hubungan yang belum diketahui. MARS dibentuk dari kombinasi regresi spline dan recursive partitioning regression. Pengembangan metode MARS dengan Bivariate Generalized Poisson Regression (BGPR) menjadi satu model yaitu Multivariate Adaptive Bivariate Generalized Poisson Regression Spline (MABGPRS). Model MABGPRS dapat memodelkan dua variabel respon dengan banyak variabel prediktor dan dengan pola hubungan yang belum diketahui. Penaksiran parameter pada model ini menggunakan metode Maximum Likelihood Estimation (MLE) dan dengan pengujian hipotesis Maximum Likelihood Ratio Test (MLRT). Model MABGPRS diterapkan ke data jumlah kasus stunting dan wasting pada Provinsi Sulawesi Tenggara pada Tahun 2021. Sebanyak 11 variabel prediktor digunakan dan didapatkan bahwa sembilan variabel prediktor yang berbeda mempengaruhi variabel respon jumlah kasus stunting dan jumlah kasus wasting. Variabel persentase anak umur 12 - 59 bulan memiliki tingkat kepentingan yang tinggi untuk model MABGPRS pada variabel jumlah kasus stunting. Variabel persentase balita yang menderita diare memiliki tingkat kepentingan yang paling tinggi untuk model MABGPRS pada variabel jumlah kasus wasting. Variabel persentase balita mendapat Vitamin A dan persentase ibu nifas mendapat Vitamin A, tidak berkontribusi terhadap pembentukan model MABGPRS untuk jumlah kasus stunting. Variabel persentase ibu nifas mendapat Vitamin A dan persentase kunjungan ibu hamil tidak berkontribusi terhadap pembentukan model MABGPRS untuk jumlah kasus wasting.
=====================================================================================================================================
Multivariate Adaptive Regression Spline (MARS) is a non-parametric regression that can handle data with many predictor variables. The MARS method can be applied to data with unknown relationship patterns. MARS is formed from a combination of spline regression and recursive partitioning regression. The development of the MARS method with Bivariate Generalized Poisson Regression (BGPR) into one model, namely the Multivariate Adaptive Bivariate Generalized Poisson Regression Spline (MABGPRS). The MABGPRS model can model two response variables with many predictor variables and with unknown relationship patterns. Parameter estimation in this model uses the Maximum Likelihood Estimation (MLE) method and with the Maximum Likelihood Ratio Test (MLRT) hypothesis testing. The MABGPRS model was applied to data on the number of stunting and wasting cases in Southeast Sulawesi Province in 2021. A total of 11 predictor variables were used, and it was found that nine different predictor variables affected the response variables of the number of stunting cases and the number of wasting cases. The variable percentage of children aged 12-59 months has a high level of importance for the MABGPRS model on the variable number of stunting cases. The variable percentage of toddlers suffering from diarrhea has the highest level of importance for the MABGPRS model on the variable number of wasting cases. The variables percentage of toddlers receiving Vitamin A and the percentage of postpartum mothers receiving Vitamin A do not contribute to the formation of the MABGPRS model for the number of stunting cases. The variables percentage of postpartum mothers receiving Vitamin A and the percentage of visits by pregnant women do not contribute to the formation of the MABGPRS model for the number of wasting cases.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Multivariate Adaptive Regression Spline, Multivariate Adaptive Bivariate Generalized Poisson Regression Spline, stunting, wasting, Maximum Likelihood Estimation.
Subjects:	Q Science > QA Mathematics > QA371 Differential equations--Numerical solutions Q Science > QA Mathematics > QA278 Cluster Analysis. Multivariate analysis. Correspondence analysis (Statistics)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis
Depositing User:	Tharatya Kartika Sadhana
Date Deposited:	10 Aug 2024 15:19
Last Modified:	26 Aug 2024 02:45
URI:	http://repository.its.ac.id/id/eprint/114806

Actions (login required)

View Item