Heryanto, Adam Zulfi (2026) Perbandingan Metode LASSO Regularized Logistic Regression. Binary Mixed Model Forest, Dan SVM Dalam Klasifikasi Risiko Konsumsi Biaya Kesehatan Peserta BPJS Kesehatan Di Jawa Timur. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5006221065-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5006221065-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy |
Abstract
Peningkatan pemanfaatan layanan kesehatan peserta BPJS Kesehatan berimplikasi pada meningkatnya biaya pelayanan kesehatan yang harus ditanggung oleh sistem Jaminan Kesehatan Nasional (JKN). Kondisi ini mendorong perlunya identifikasi peserta yang berisiko tinggi dalam mengonsumsi layanan kesehatan sebagai upaya mendukung pengelolaan risiko dan keberlanjutan pembiayaan kesehatan. Penelitian ini bertujuan menganalisis dan membandingkan kinerja metode LASSO Regularized Logistic Regression, Binary Mixed Model (BiMM) Forest, dan Support Vector Machine (SVM) dalam mengklasifikasikan risiko konsumsi layanan kesehatan peserta BPJS Kesehatan di Provinsi Jawa Timur tahun 2024. Penelitian menggunakan data sekunder BPJS Kesehatan tahun 2024 sebanyak 206.702 observasi. Variabel respon yang digunakan adalah status klaim tinggi, sedangkan variabel prediktor meliputi usia peserta, jenis kelamin, kelas rawat, jenis fasilitas kesehatan, tipe fasilitas kesehatan, tingkat pelayanan, jenis poli, special procedures, dan special drugs. Pada metode LASSO dilakukan seleksi variabel menggunakan regularisasi L1, pada BiMM Forest dilakukan seleksi variabel menggunakan VarSeIRF dan VSURF, sedangkan pada SVM digunakan kernel linear, polynomial, dan Radial Basis Function (RBF). Kinerja model dievaluasi menggunakan accuracy, sensitivity, specificity, precision, F1-score, dan Area Under the Curve (AUC). Hasil penelitian menunjukkan model terbaik diperoleh pada Random Forest dengan seleksi variabel VSURF tanpa mempertimbangkan mixed effect yang menghasilkan accuracy sebesar 0,8543, specificity sebesar 0,9728, precision sebesar 0,7671, F1-score sebesar 0,4972, dan AUC sebesar 0,8040. Model LASSO Regularized Logistic Regression menghasilkan accuracy sebesar 0,8054 dan AUC sebesar 0,7356, sedangkan model SVM kernel Polynomial memperoleh accuracy sebesar 0,7938 dan AUC sebesar 0,7568. Berdasarkan hasil evaluasi, Random Forest dengan seleksi variabel VSURF tanpa mixed effect merupakan metode yang paling optimal untuk mengklasifikasikan risiko konsumsi layanan kesehatan peserta BPJS Kesehatan di Provinsi Jawa Timur tahun 2024
===========================================================================================================================================
The increasing utilization of healthcare services by BPJS Kesehatan participants has led to a substantial rise in healthcare expenditures covered by Indonesia’s National Health Insurance (JKN) system. This condition highlights the need to identify participants with a high risk of healthcare service utilization in order to support risk management and maintain the sustainability of healthcare financing. This study aims to analyze and compare the performance of LASSO Regularized Logistic Regression, Binary Mixed Model (BiMM) Forest, and Support Vector Machine (SVM) in classifying the risk of healthcare service utilization among BPJS Kesehatan participants in East Java Province in 2024. This study utilized secondary data from BPJS Kesehatan in 2024 consisting of 206,702 observations. The response variable was high-claim status, while the predictor variables included participant age, gender, inpatient class, healthcare facility type, healthcare facility category, service level, polyclinic type, special procedures, and special drugs. In the LASSO model, variable selection was performed using L1 regularization. For the BiMM Forest model, variable selection was conducted using VarSeIRF and VSURF, while the SVM model employed linear, polynomial, and Radial Basis Function (RBF) kernels. Model performance was evaluated using accuracy, sensitivity, specificity, precision, F1-score, and Area Under the Curve (AUC). The results indicate that the best-performing model was Random Forest with VSURF variable selection without incorporating mixed effects, achieving an accuracy of 0.8543, specificity of 0.9728, precision of 0.7671, F1-score of 0.4972, and AUC of 0.8040. LASSO Regularized Logistic Regression achieved an accuracy of 0.8054 and an AUC of 0.7356, while the SVM model with a polynomial kernel achieved an accuracy of 0.7938 and an AUC of 0.7568. Based on the evaluation results, Random Forest with VSURF variable selection without mixed effects was found to be the most effective method for classifying the risk of healthcare service utilization among BPJS Kesehatan participants in East Java Province in 2024
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | BiMM Forest, binary classification, BPJS Kesehatan, LASSO logistic regression, Support Vector Machine, klasifikasi biner, LASSO logistic regression, Support Vector Machine. |
| Subjects: | H Social Sciences > HA Statistics > HA31.3 Regression. Correlation. Logistic regression analysis. Q Science Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression Q Science > QA Mathematics > QA353.K47 Kernel functions (analysis) T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models. T Technology > T Technology (General) > T58.62 Decision support systems |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis |
| Depositing User: | Adam Zulfi Heryanto |
| Date Deposited: | 17 Jul 2026 07:36 |
| Last Modified: | 17 Jul 2026 07:36 |
| URI: | http://repository.its.ac.id/id/eprint/135312 |
Actions (login required)
 |
View Item |