Show simple item record

dc.contributor.advisorSetiawaty, Berlian
dc.contributor.authorDevi, Ishwari Chandani
dc.date.accessioned2026-08-03T06:48:59Z
dc.date.available2026-08-03T06:48:59Z
dc.date.issued2026
dc.identifier.urihttp://repository.ipb.ac.id/handle/123456789/176860
dc.description.abstractSebaran data klaim asuransi kebakaran umumnya berekor berat (heavy-tailed), didominasi klaim kecil yang sering terjadi dan klaim ekstrem yang jarang namun berdampak besar, sehingga sebaran tunggal sering tidak memadai untuk memodelkannya. Penelitian ini bertujuan membangun model sebaran splicing untuk memodelkan besar klaim asuransi serta mengukur dan membandingkan risikonya menggunakan Value at Risk (VaR) dan Tail Value at Risk (TVaR). Data yang digunakan adalah klaim asuransi kebakaran Denmark sebanyak 2.167 observasi. Model sebaran splicing menggabungkan sebaran lognormal terpancung pada bagian body dan sebaran Pareto II pada bagian tail dengan threshold sebesar 5 juta Danish Krone (DKK). Estimasi parameter dilakukan dengan metode maximum likelihood, sedangkan VaR dan TVaR dihitung secara analitik dan simulasi. Hasil uji goodness-of-fit menunjukkan model sebaran splicing memberikan kecocokan terbaik dibandingkan model sebaran tunggal lognormal dan Pareto II. Pada tingkat kepercayaan 95%, diperoleh VaR sebesar 9,299 juta DKK dan TVaR sebesar 26,986 juta DKK, sedangkan pada tingkat kepercayaan 99%, diperoleh VaR sebesar 27,506 juta DKK dan TVaR sebesar 76,355 juta DKK.
dc.description.abstractFire insurance claim data distributions are generally heavy-tailed, dominated by frequent small claims accompanied by rare but highly impactful extreme claims, so that a single distribution is often inadequate to model them. This study aims to develop a spliced distribution model for insurance claim severity and to measure and compare its risk using Value at Risk (VaR) and Tail Value at Risk (TVaR). The data used are Danish fire insurance claims consisting of 2,167 observations. The spliced distribution model combines a truncated lognormal distribution for the body and a Pareto II distribution for the tail, with a threshold of 5 million Danish Krone (DKK). Parameters were estimated by the maximum likelihood method, while VaR and TVaR were computed analytically and by simulation. The goodness-of-fit test shows that the spliced distribution model provides the best fit compared to single lognormal and Pareto II distributions. At the 95% confidence level, the VaR is 9.299 million DKK and the TVaR is 26.986 million DKK, while at the 99% confidence level, the VaR is 27.506 million DKK and the TVaR is 76.355 million DKK.
dc.description.sponsorship
dc.language.isoid
dc.publisherIPB Universityid
dc.titleAnalisis Value at Risk dan Tail Value at Risk pada Data Besar Klaim Asuransi dengan Model Sebaran Splicingid
dc.title.alternativeAnalysis of Value at Risk and Tail Value at Risk on the Amount of Insurance Claims Data Using a Spliced Distribution Model
dc.typeSkripsi
dc.subject.keywordheavy-tailedid
dc.subject.keywordmodel sebaran splicingid
dc.subject.keywordPareto IIid
dc.subject.keywordTail Value at Risk (TVaR)id
dc.subject.keywordValue at Risk (VaR)id
dc.subject.keywordspliced distribution modelid
dc.subtypeUndergraduate Theses


Files in this item

Thumbnail
Thumbnail
Thumbnail

This item appears in the following Collection(s)

Show simple item record