Pemodelan Status Perceraian Wanita Usia Subur Di Jawa Barat Menggunakan Metode Regresi Logistik Biner Smote (Synthetic Minority Over-Sampling Technique)

Fitriani, Mellina Eka (2023) Pemodelan Status Perceraian Wanita Usia Subur Di Jawa Barat Menggunakan Metode Regresi Logistik Biner Smote (Synthetic Minority Over-Sampling Technique). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 06211940000099_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (5MB) | Request a copy

Abstract

Perceraian merupakan permasalahan sosial yang tertinggi di Provinsi Jawa Barat dibandingkan tingkat nasional. Pada tahun 2022 terdapat 113.643 kasus perceraian yang terjadi, seiring meningkatnya kompensasi pernikahan usia dini. Variabel dependen yang digunakan pada penelitian ini merupakan data kategori berskala biner yaitu cerai dan tidak cerai sehingga metode yang digunakan yaitu regresi logistik biner. Terdapat 10 variabel independen yaitu usia menikah pertama, lama pernikahan, jumlah anak, geografis tempat tinggal, pendidikan terakhir, pekerjaan, kepemilikan jaminan kesehatan, kehamilan tidak diinginkan, pengetahuan pembangunan keluarga, dan akses informasi. Data yang digunakan adalah hasil Survei Kinerja Akuntabilitas Program (SKAP) 2019 dengan satuan observasi Wanita Usia Subur (WUS) usia 10-49 tahun. WUS diidentifikasi bercerai apabila memiliki status cerai hidup. Persentase WUS bercerai hanya 17,2% dari 2.161 WUS mengindikasikan data tidak seimbang, sehingga ditangani dengan SMOTE. Model regresi logistik biner memiliki nilai accuracy 0,832, APER 0,167, sensitivity 0,986, spesificity 0,088, precission 0,839 dan AUC 0,537, sedangkan model regresi logistik setelah SMOTE memiliki nilai accuracy 0,715, APER 0,246, sensitivity 0,768, specificity 0,649, precission 0,732 dan AUC 0,708. Model regresi logistik biner setelah SMOTE sudah lebih baik dalam memprediksi kelas minoritas yaitu meningkatkan nilai specificity dari 0.088 menjadi 0.649 serta nilai AUC dari 0,537 menjadi 0,708. Meskipun nilai accuracy dan sensitivity model regresi logistik biner dengan SMOTE lebih rendah namun tidak mendekati nol seperti specificity pada model regresi logistik biner. Hasil perbandingan juga menunjukkan analisis regresi logistik biner setelah SMOTE membuat lebih banyak variabel yang signifikan berpengaruh dibandingkan tanpa SMOTE. Variabel yang signifikan yaitu variabel usia menikah pertama, lama pernikahan, jumlah anak, pendidikan terakhir, kepemilikan jaminan kesehatan dan akses informasi.
=================================================================================================================================
Divorce is the most serious social problem in West Java Province when compared to the national average. There were 113.643 divorce cases in 2022, with increasing compensation for early marriages. The dependent variable in this study is binary-scale category data, namely divorced and not divorced, so binary logistic regression was used. Age at first marriage, length of marriage, number of children, geographical place of residence, last education, occupation, ownership of health insurance, unwanted pregnancies, knowledge of family development, and access to information are the ten independent variables. The results of the 2019 Program Accountability Performance Survey (SKAP) with the observation unit Women of Reproductive Age (WUS) aged 10-49 years were used. If a WUS has divorced status, they are classified as divorced. Only 17.2% of the 2.161 WUS are divorced, which indicates uneven data and the need for SMOTE. The accuracy of the binary logistic regression model is 0.832, the APER is 0.167, the sensitivity is 0.986, the specificity is 0.088, precission 0.839 and the AUC is 0.537, whereas the accuracy of the logistic regression model after SMOTE is 0.715, the APER is 0.246, the sensitivity is 0.768, the specificity is 0.649, the precission 0,732 and the AUC is 0.708. Following SMOTE, the binary logistic regression model performs better in terms of predicting the minority class, with specificity values rising from 0.088 to 0.649 and AUC values rising from 0.537 to 0.708. However, unlike the specificity of the binary logistic regression model, the accuracy and sensitivity values of the binary logistic regression model with SMOTE are not near to zero. The comparison results show that binary logistic regression analysis with SMOTE significantly influences more variables than without SMOTE. The significant variables are age at first marriage, length of marriage, number of children, last education, ownership of health insurance and access to information.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Data Imbalanced, Perceraian, Regresi Logistik Biner, SMOTE, Wanita Usia Subur (WUS), Binary Logistic Regression, Divorce, Imbalanced Data, Women of Reproductive Age (WUS)
Subjects:	H Social Sciences > HA Statistics > HA29 Theory and method of social science statistics H Social Sciences > HQ The family. Marriage. Woman 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:	Mellina Eka Fitriani
Date Deposited:	17 Oct 2023 07:49
Last Modified:	17 Oct 2023 07:49
URI:	http://repository.its.ac.id/id/eprint/103532

Actions (login required)

View Item