Banudwaya, Pragnya (2026) Evaluasi Poisson, Fractional Poisson, dan Mittag-Leffler Count Model Dalam Perbandingan Tingkat Premi dengan Kredibilitas Bühlmann. Other thesis, Institut Teknologi Sepuluh Nopember.
|
Text
5006221039_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy |
Abstract
Risiko merupakan bagian yang tidak terpisahkan dari aktivitas manusia, khususnya bagi perusahaan asuransi sebagai pihak ketiga dalam mekanisme pengalihan risiko. Data dari Asosiasi Asuransi Umum Indonesia mencatat rasio klaim sebesar 44,3% pada tahun 2023 dan 44,2% pada tahun 2024, sementara SEOJK Nomor 6 Tahun 2017 membatasi penyesuaian premi yang diperbolehkan. Kondisi ini menegaskan pentingnya kemampuan perusahaan asuransi dalam menetapkan dan menyesuaikan premi secara akurat agar terhindar dari potensi kerugian besar. Salah satu pendekatan dalam perhitungan premi adalah teori kredibilitas, yang menyesuaikan premi masa depan berdasarkan pengalaman historis suatu risiko individu maupun kelompok risiko. Pendekatan yang paling banyak digunakan adalah model Bühlmann dan Bühlmann-Straub, keduanya bergantung pada asumsi distribusi tertentu. Dalam estimasi frekuensi klaim, Poisson count model umumnya digunakan. Akan tetapi, asumsi equidispersion pada model ini membatasi penerapannya, sehingga diperlukan model alternatif seperti fractional Poisson dan Mittag-Leffler count model yang lebih sesuai untuk data dengan underdispersion maupun overdispersion. Penelitian ini bertujuan untuk membandingkan Poisson, fractional Poisson, dan Mittag-Leffler count model dalam perhitungan kredibilitas Bühlmann untuk penyesuaian premi asuransi kendaraan tahun 2024, di mana asumsi equidispersion tidak terpenuhi. Kedua model alternatif dipilih karena kemampuannya dalam menangani data non-equidispersed serta kesamaan distribusi interarrival times yang mengikuti fungsi Mittag-Leffler, sehingga memungkinkan perbandingan yang adil. Evaluasi kinerja model dilakukan menggunakan kriteria AIC dan BIC. Hasil penelitian menunjukkan bahwa Poisson count model memiliki nilai AIC dan BIC terendah dibandingkan model lainnya. Akan tetapi, dalam perhitungan premi kredibilitas, Poisson dan fractional Poisson count model menghasilkan faktor kredibilitas mendekati 1, yang menyebabkan penyesuaian premi menjadi sangat ekstrem dan kurang proporsional terhadap risiko kolektif. Sebaliknya, Mittag-Leffler count model menghasilkan faktor kredibilitas sebesar 0,6010, yang memberikan bobot penyesuaian yang lebih seimbang. Hasil perbandingan tingkat premi menunjukkan bahwa Mittag-Leffler count model menghasilkan penyesuaian premi yang lebih seimbang, realistis, dan adil bagi pemegang polis dibandingkan dua model lainnya. Oleh karena itu, Mittag-Leffler count model direkomendasikan sebagai model alternatif yang lebih baik dalam menjaga keberlanjutan penetapan premi asuransi.
======================================================================================================================================
Risk is an inseparable part of human activity, particularly for insurance companies as third parties in risk transfer. Data from the Indonesian General Insurance Association recorded claim ratios of 44.3% in 2023 and 44.2% in 2024, while SEOJK Regulation No. 6 of 2017 restricts permissible premium adjustments. This underscores the importance of insurers’ ability to set and adjust premiums accurately to avoid large losses. One approach to premium calculation is credibility theory, which adjusts future premiums based on the historical experience of an individual risk or group of risks. The most widely applied approaches are the Bühlmann and Bühlmann-Straub models, both of which rely on distributional assumptions. In estimating claim frequency, the Poisson count model is commonly used. However, its equidispersion assumption limits applicability, making alternative models such as the fractional Poisson and Mittag-Leffler count models more suitable for data with underdispersion or overdispersion. This study aims to compare the Poisson, fractional Poisson, and Mittag-Leffler count models in Bühlmann credibility calculations for motor insurance premium adjustment in 2024, where the equidispersion assumption is not met. The two alternative models are chosen for their ability to handle nonequidispersed data and their shared interarrival time distribution, namely the Mittag-Leffler function, allowing for a fair comparison. Model performance is evaluated using AIC and BIC. The research results showed that the Poisson count model had the lowest AIC and BIC values compared to the other models. However, in the credibility premium calculation, the Poisson and fractional Poisson count models produced credibility factors close to 1, causing premium adjustments to become very extreme and less proportional to the collective risk. In contrast, the Mittag-Leffler count model produced a credibility factor of 0.6010, providing a more balanced adjustment weight. The results of the premium level comparison showed that the Mittag-Leffler count model produced premium adjustments that were more balanced, realistic, and fair for policyholders compared to the other two models. Therefore, the Mittag-Leffler count model is recommended as a better alternative model in maintaining the sustainability of insurance premium setting.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Bühlmann, Fractional Poisson count model, Mittag-Leffler count model, Poisson count model, Premi Bühlmann, Fractional Poisson count model, Mittag-Leffler count model, Poisson count model, Premium |
| Subjects: | Q Science > QA Mathematics > QA275 Theory of errors. Least squares. Including statistical inference. Error analysis (Mathematics) Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA279.5 Bayesian statistical decision theory. Q Science > QA Mathematics > QA402.5 Genetic algorithms. Interior-point methods. |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis |
| Depositing User: | Pragnya Banudwaya |
| Date Deposited: | 12 Jan 2026 05:43 |
| Last Modified: | 12 Jan 2026 05:43 |
| URI: | http://repository.its.ac.id/id/eprint/129492 |
Actions (login required)
![]() |
View Item |
