Masita, Aisha Rahma Putri (2024) Analisis Perbandingan Metode Decision Tree Regression Dan Random Forest Regression Pada Prediksi Kepesertaan Jaminan Kesehatan Nasional. Other thesis, Institut Teknologi Sepuluh Nopember.
Text
5006211002-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (4MB) | Request a copy |
Abstract
Kesehatan adalah salah satu aspek penting dalam mencapai kesejahteraan individu dan keberhasilan pembangunan negara. Untuk menjamin akses layanan kesehatan yang merata, pemerintah Indonesia meluncurkan program Jaminan Kesehatan Nasional (JKN) pada 1 Januari 2014. Namun, jumlah peserta aktif JKN hanya mencapai sekitar 80% pada tahun 2024, yang menyebabkan defisit keuangan pada BPJS Kesehatan. Penelitian ini bertujuan untuk memprediksi jumlah peserta aktif JKN menggunakan metode Decision Tree Regression dan Random Forest Regression, serta membandingkan performa kedua metode tersebut. Variabel respon adalah jumlah peserta aktif JKN, sedangkan variabel independen berupa jumlah fasilitas kesehatan seperti jumlah puskesmas, jumlah dokter praktik perorangan, jumlah klinik pratama termasuk klinik TNI/POLRI, jumlah FKRTL, jumlah penduduk miskin, tingkat pengangguran terbuka, pengeluaran perkapita disesuaikan, jumlah penduduk, angka harapan hidup (AHH), dan rata-rata konsumsi non makanan rumah tangga yang memiliki pengeluaran telekomunikasi, pada periode data dari 2016 – 2021. Hasil penelitian menunjukkan bahwa jumlah FKRTL, jumlah penduduk, jumlah dokter, dan jumlah puskesmas secara konsisten memberikan kontribusi besar dalam prediksi pada metode Decision Tree Regression maupun Random Forest Regression. Selain itu, Decision Tree Regression mengidentifikasi jumlah klinik pratama dan jumlah penduduk miskin sebagai variabel tambahan yang berpengaruh. Decision Tree Regression menghasilkan nilai MAE sebesar 772.464,5, nilai MAPE sebesar 27,80%, dan nilai RMSE sebesar 1.324.906. Random Forest Regression memberikan hasil lebih baik dengan nilai MAE sebesar 518.909,8, nilai MAPE sebesar 14,79%, dan nilai RMSE sebesar 923.888,9. Pengujian dengan preprocessing standardized memberikan hasil serupa. Secara keseluruhan, Random Forest Regression terbukti lebih baik dalam akurasi prediksi terlihat dari nilai MAE, MAPE, dan RMSE yang lebih kecil dibandingkan dengan Decision Tree Regression. Hal ini karena kemampuannya mengurangi kesalahan dengan menggabungkan prediksi dari banyak tree. Penelitian ini memberikan wawasan penting bagi BPJS Kesehatan untuk mengembangkan strategi berbasis data dan optimalisasi fasilitas kesehatan sebagai strategi utama dalam meningkatkan keikutsertaan masyarakat pada program JKN.
=================================================================================================================================
Health is one of the important aspects in achieving individual welfare and the success of national development. To ensure equal access to health services, the Indonesian government launched the National Health Insurance (JKN) program on January 1, 2014. However, the number of active JKN participants will only reach around 80% in 2024, which will cause a financial deficit in BPJS Kesehatan. This study aims to predict the number of active JKN participants using the Decision Tree Regression and Random Forest Regression methods, and to compare the performance of the two methods. The response variable is the number of active JKN participants, while the independent variables are the number of health facilities such as the number of health centers, the number of individual practicing doctors, the number of primary clinics including TNI/POLRI clinics, the number of FKRTL, the number of poor people, the open unemployment rate, adjusted per capita expenditure, population, life expectancy (AHH), and the average non-food consumption of households that have telecommunications expenditure, in the data period from 2016 - 2021. The results of the study showed that the number of FKRTL, population, number of doctors, and number of health centers consistently made a large contribution to the prediction of the Decision Tree Regression and Random Forest Regression methods. In addition, Decision Tree Regression identified the number of primary clinics and the number of poor people as additional influential variables. Decision Tree Regression produced a MAE value of 772,464.5, a MAPE value of 27.80%, and an RMSE value of 1,324,906. Random Forest Regression gives better results with MAE value of 518,909.8, MAPE value of 14.79%, and RMSE value of 923,888.9. Testing with standardized preprocessing gives similar results. Overall, Random Forest Regression is proven to be better in prediction accuracy as seen from the smaller MAE, MAPE, and RMSE values compared to Decision Tree Regression. This is due to its ability to reduce errors by combining predictions from many trees. This study provides important insights for BPJS Kesehatan to develop databased strategies and optimize health facilities as the main strategy in increasing community participation in the JKN program.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Prediction, National Health Insurance, Participation Decision Tree Regression, Random Forest Regression, Prediksi, Jaminan Kesehatan Nasional, Kepesertaan |
Subjects: | Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression |
Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis |
Depositing User: | Aisha Rahma Putri Masita |
Date Deposited: | 03 Jan 2025 07:40 |
Last Modified: | 03 Jan 2025 07:40 |
URI: | http://repository.its.ac.id/id/eprint/116104 |
Actions (login required)
View Item |