Pemodelan Faktor-Faktor yang Memengaruhi Status Perceraian Wanita Usia Subur Menggunakan Rare Event Weighted Logistic Regression (RE-WLR)

Najah, Nur Kholifahtus Jannatun (2025) Pemodelan Faktor-Faktor yang Memengaruhi Status Perceraian Wanita Usia Subur Menggunakan Rare Event Weighted Logistic Regression (RE-WLR). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003211103-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Perceraian adalah putusnya hubungan perkawinan antara suami-istri berdasarkan putusan pengadilan. Perceraian yang terjadi dapat dilihat dari status perkawinan salah satunya pada wanita usia subur (WUS) yang telah menikah. Definisi WUS adalah wanita berusia 15-49 tahun dengan organ reproduksi yang berfungsi optimal sehingga mampu hamil dan melahirkan. Dalam suatu survei berskala nasional, status perkawinan WUS diklasifikasikan dalam empat kategori yaitu belum kawin, kawin, cerai hidup, dan cerai mati. Penelitian ini hanya berfokus pada WUS dengan status perkawinan nikah dan cerai hidup. Status tersebut dikategorikan ulang menjadi tidak cerai dan cerai sebagai variabel respon yang disebut status perceraian. Berdasarkan data Survei Sosial Ekonomi Nasional (SUSENAS) tahun 2023, di Provinsi Jawa Timur terdapat 12.220 sampel individu dengan 96,67% tidak cerai dan 3,33% cerai. Data tersebut menunjukkan adanya ketidakseimbangan data (imbalanced) yang sangat jauh dengan jumlah data berskala besar. Pada penelitian ini, data dibagi menjadi data training dan data testing untuk mengevaluasi performa model sehingga berpotensi terjadi overfitting pada model yang dilatih. Oleh karena itu, penelitian ini menggunakan metode Rare Event Weighted Logistic Regression (RE-WLR) yang menerapkan regularisasi serta pembobotan untuk menangani imbalanced data dan mengurangi overfitting. Data yang digunakan merupakan data SUSENAS KOR Jawa Timur 2023 dengan 1 variabel respon dan 10 variabel prediktor. Hasil dari penelitian ini menunjukkan dari 10 variabel prediktor yang digunakan terdapat 7 variabel yang berpengaruh signifikan terhadap model yaitu usia kawin pertama (X_2), status pekerjaan (X_4), kepemilikan anak (X_5), tempat tinggal (X_6), kepemilikan jaminan kesehatan (X_7), kepemilikan sumber penghasilan (X_8), dan gangguan emosional (X_9). Ketepatan model dalam mengklasifikasikan status perceraian di Provinsi Jawa Timur menghasilkan akurasi sebesar 96,69%, sensitivitas 4,94%, spesifisitas 99,83%, G-Mean 22,20%, dan AUC sebesar 52,38%.
======================================================================================================================================
Divorce is the breakup of marital relations between husband and wife based on a court decision. Divorce that occurs can be seen from the marital status of one of them in women of childbearing age (WUS) who have married. The definition of WUS is a woman aged 15-49 years with reproductive organs that function optimally so that she is able to become pregnant and give birth. In a national survey, the marital status of WUS was classified into four categories: unmarried, married, divorced, and widowed. This research focused only on WUS with married and divorced marital status. These statuses were re-categorized into not divorced and divorced as a response variable called divorce status. Based on data from the National Socio-Economic Survey (SUSENAS) in 2023, in East Java Province there were 11,930 individual samples with 96.67% not divorced and 3.33% divorced. The data shows a very imbalanced data with a large amount of data. In this research, the dataset is partitioned into training and testing subsets to evaluate the model’s performance, which may increase the risk of overfitting in the trained model. Therefore, this research applies the Rare Event Weighted Logistic Regression (RE-WLR) method, which incorporates regularization and weighting schemes to address data imbalance and mitigate the risk of overfitting. The data used is East Java SUSENAS data with 1 response variable and 10 predictor variables. The results of this research indicate that of the 10 predictor variables used, 7 variables have a significant effect on the model, namely age at first marriage (X₂), employment status (X₄), having children (X₅), place of residence (X₆), health insurance ownership (X₇), having a source of income (X₈), and emotional disorders (X₉). The model’s accuracy in classifying divorce status in East Java Province is 96,69%, with a sensitivity of 4,94%, specificity of 99,83%, G-Mean of 22,20%, and an AUC of 52,38%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Imbalanced Data, RE-WLR, Status Perceraian, Wanita Usia Subur, Divorce Status, Imbalanced data, RE-WLR, Women of Childbearing Age
Subjects:	Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Nur Kholifahtus Jannatun Najah
Date Deposited:	03 Aug 2025 06:17
Last Modified:	03 Aug 2025 06:17
URI:	http://repository.its.ac.id/id/eprint/125493

Actions (login required)

View Item