Parameter: Jurnal Matematika, Statistika dan Terapannya
Vol 4 No 3 (2025): Parameter: Jurnal Matematika, Statistika dan Terapannya

Comparison of Random Forest and XGBoost Methods for Predicting Work Accident Claim Reserves

Anugrah, Sri Ayu (Unknown)
Anugrawati, Sri Dewi (Unknown)
Sauddin, Adnan (Unknown)
Mariani, Andi (Unknown)



Article Info

Publish Date
23 Dec 2025

Abstract

The potential high claim burden in the work accident insurance sector managed by BPJS Ketenagakerjaan have an impact on the company’s financial stability. This encourages insurance companies to provide additional funds to maintain the company’s operational sustainability. Thus, preparing future fund reserves is a crucial step in risk and financial management to minimize payment delays, up to the risk of default. This study aims to determine the best method for predicting work accident claim reserves by comparing the Random Forest and XGBoost methods. The result of the analysis shows that the XGBoost method has an outstanding ability to predict work accident claim reserves on BPJS Ketenagakerjaan in the period July 2016 to August 2023, with a MAPE of 5.14% and an accuracy rate of 94.86%.

Copyrights © 2025






Journal Info

Abbrev

parameter

Publisher

Subject

Mathematics

Description

Parameter: Jurnal Matematika, Statistika dan Terapannya is an open access journal (e-journal) published since April 2022. Parameteris published by Department of Mathematics, Faculty of Science and Mathematics, Pattimura. Parameterpublished scientific articles on various aspects related to ...