{"id":{"repo_id":"vilnius","oai_identifier":"oai:vu.lt:elaba:308334579"},"canonical_url":"https://search.ndltd.org/etd/vilnius/oai:vu.lt:elaba:308334579","repository":{"repo_id":"vilnius","name":"Vilnius University","base_url":"https://epublications.vu.lt/oai"},"display":{"title":"Motorinių transporto priemonių draudimo įmokų ir žalų rizikos analizė naudojant statistinius ir mašininio mokymosi metodus /","abstract":"This bachelor’s thesis analyses motor vehicle insurance premiums and claim occurrence risk using statistical and machine learning methods. The empirical study is based on a real anonymi&#8208; sed motor vehicle insurance portfolio containing contract&#8208;related variables, customer history, claim information and technical vehicle characteristics. Since the same ID may appear in several annual records, the train&#8208;test split and cross&#8208;validation were performed at the ID level in order to reduce the risk of information leakage between training and test samples. Two models were developed for premium prediction: a Gamma generalized linear model and an XGBoost regression model. The Gamma generalized linear model was used as an interpretable statistical model suitable for positive and right&#8208;skewed premiums, while XGBoost was used as a more flexible method capable of capturing nonlinear relationships and feature interactions. In the test sample, XGBoost achieved lower prediction errors than the Gamma generalized linear model, with MAE equal to 66.62 and RMSE equal to 102.17, compared with 70.97 and 109.21 for the Gamma generalized linear model. For claim occurrence prediction, a binomial generalized linear model and an XGBoost classifi&#8208; cation model were estimated. Both models improved on the baseline model, although their discri&#8208; mination ability was moderate. XGBoost achieved slightly better performance, with ROC AUC equal to 0.6489, PR AUC equal to 0.2953 and Brier score equal to 0.1433. Customer segmentation was additionally performed using PCA and k&#8208;means clustering. Three practically interpretable segments were obtained: more active customers with larger vehicles, mo&#8208; torcycle customers and the main passenger car customer segment. Since the silhouette coefficient was not high, the resulting segments should be interpreted as practical portfolio groups rather than strictly separated natural customer classes.","abstract_html":"This bachelor’s thesis analyses motor vehicle insurance premiums and claim occurrence risk using statistical and machine learning methods. The empirical study is based on a real anonymi&amp;#8208; sed motor vehicle insurance portfolio containing contract&amp;#8208;related variables, customer history, claim information and technical vehicle characteristics. Since the same ID may appear in several annual records, the train&amp;#8208;test split and cross&amp;#8208;validation were performed at the ID level in order to reduce the risk of information leakage between training and test samples. Two models were developed for premium prediction: a Gamma generalized linear model and an XGBoost regression model. The Gamma generalized linear model was used as an interpretable statistical model suitable for positive and right&amp;#8208;skewed premiums, while XGBoost was used as a more flexible method capable of capturing nonlinear relationships and feature interactions. In the test sample, XGBoost achieved lower prediction errors than the Gamma generalized linear model, with MAE equal to 66.62 and RMSE equal to 102.17, compared with 70.97 and 109.21 for the Gamma generalized linear model. For claim occurrence prediction, a binomial generalized linear model and an XGBoost classifi&amp;#8208; cation model were estimated. Both models improved on the baseline model, although their discri&amp;#8208; mination ability was moderate. XGBoost achieved slightly better performance, with ROC AUC equal to 0.6489, PR AUC equal to 0.2953 and Brier score equal to 0.1433. Customer segmentation was additionally performed using PCA and k&amp;#8208;means clustering. Three practically interpretable segments were obtained: more active customers with larger vehicles, mo&amp;#8208; torcycle customers and the main passenger car customer segment. Since the silhouette coefficient was not high, the resulting segments should be interpreted as practical portfolio groups rather than strictly separated natural customer classes.","abstract_has_math":false,"creators":["Jermakovičius, Ignas,"],"institution":"Institutional Repository of Vilnius University","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-19T10:12:55Z","subjects":[],"languages":["lit"],"rights":["info:eu-repo/semantics/openAccess"],"rights_urls":[],"identifier_entries":[],"person_identifier_entries":[]},"links":{"outbound_url":"https://repository.vu.lt/VU:ELABAETD308334579&prefLang=en_US","outbound_label":"Repository record","outbound_source":"dc:identifier","record_url":"https://repository.vu.lt/VU:ELABAETD308334579&prefLang=en_US","record_label":"Repository record","record_kind":"record","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":"open","embargo_until":null,"access_note":null},"source_record":{"url":"https://epublications.vu.lt/oai?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Avu.lt%3Aelaba%3A308334579","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Jermakovičius, Ignas,"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2026"]},{"key":"dc:publisher","label":"Institution","values":["Institutional Repository of Vilnius University"]},{"key":"dc:relation","label":"Dc Relation","values":["https://epublications.vu.lt/object/elaba:308334579/308334579.pdf"]},{"key":"dc:type","label":"Dc Type","values":["info:eu-repo/semantics/bachelorThesis"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["lit"]},{"key":"dc:rights","label":"Dc Rights","values":["info:eu-repo/semantics/openAccess"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://repository.vu.lt/VU:ELABAETD308334579&prefLang=en_US"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["This bachelor’s thesis analyses motor vehicle insurance premiums and claim occurrence risk using statistical and machine learning methods. The empirical study is based on a real anonymi&#8208; sed motor vehicle insurance portfolio containing contract&#8208;related variables, customer history, claim information and technical vehicle characteristics. Since the same ID may appear in several annual records, the train&#8208;test split and cross&#8208;validation were performed at the ID level in order to reduce the risk of information leakage between training and test samples. Two models were developed for premium prediction: a Gamma generalized linear model and an XGBoost regression model. The Gamma generalized linear model was used as an interpretable statistical model suitable for positive and right&#8208;skewed premiums, while XGBoost was used as a more flexible method capable of capturing nonlinear relationships and feature interactions. In the test sample, XGBoost achieved lower prediction errors than the Gamma generalized linear model, with MAE equal to 66.62 and RMSE equal to 102.17, compared with 70.97 and 109.21 for the Gamma generalized linear model. For claim occurrence prediction, a binomial generalized linear model and an XGBoost classifi&#8208; cation model were estimated. Both models improved on the baseline model, although their discri&#8208; mination ability was moderate. XGBoost achieved slightly better performance, with ROC AUC equal to 0.6489, PR AUC equal to 0.2953 and Brier score equal to 0.1433. Customer segmentation was additionally performed using PCA and k&#8208;means clustering. Three practically interpretable segments were obtained: more active customers with larger vehicles, mo&#8208; torcycle customers and the main passenger car customer segment. Since the silhouette coefficient was not high, the resulting segments should be interpreted as practical portfolio groups rather than strictly separated natural customer classes."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Motorinių transporto priemonių draudimo įmokų ir žalų rizikos analizė naudojant statistinius ir mašininio mokymosi metodus /","Analysis of motor vehicle insurance premiums and claims risk using statistical and machine learning methods."]}]}],"canonical_facts":{"dc:creator":["Jermakovičius, Ignas,"],"dc:date":["2026"],"dc:description":["This bachelor’s thesis analyses motor vehicle insurance premiums and claim occurrence risk using statistical and machine learning methods. The empirical study is based on a real anonymi&#8208; sed motor vehicle insurance portfolio containing contract&#8208;related variables, customer history, claim information and technical vehicle characteristics. Since the same ID may appear in several annual records, the train&#8208;test split and cross&#8208;validation were performed at the ID level in order to reduce the risk of information leakage between training and test samples. Two models were developed for premium prediction: a Gamma generalized linear model and an XGBoost regression model. The Gamma generalized linear model was used as an interpretable statistical model suitable for positive and right&#8208;skewed premiums, while XGBoost was used as a more flexible method capable of capturing nonlinear relationships and feature interactions. In the test sample, XGBoost achieved lower prediction errors than the Gamma generalized linear model, with MAE equal to 66.62 and RMSE equal to 102.17, compared with 70.97 and 109.21 for the Gamma generalized linear model. For claim occurrence prediction, a binomial generalized linear model and an XGBoost classifi&#8208; cation model were estimated. Both models improved on the baseline model, although their discri&#8208; mination ability was moderate. XGBoost achieved slightly better performance, with ROC AUC equal to 0.6489, PR AUC equal to 0.2953 and Brier score equal to 0.1433. Customer segmentation was additionally performed using PCA and k&#8208;means clustering. Three practically interpretable segments were obtained: more active customers with larger vehicles, mo&#8208; torcycle customers and the main passenger car customer segment. Since the silhouette coefficient was not high, the resulting segments should be interpreted as practical portfolio groups rather than strictly separated natural customer classes."],"dc:format":["application/pdf"],"dc:identifier":["https://repository.vu.lt/VU:ELABAETD308334579&prefLang=en_US"],"dc:language":["lit"],"dc:publisher":["Institutional Repository of Vilnius University"],"dc:relation":["https://epublications.vu.lt/object/elaba:308334579/308334579.pdf"],"dc:rights":["info:eu-repo/semantics/openAccess"],"dc:title":["Motorinių transporto priemonių draudimo įmokų ir žalų rizikos analizė naudojant statistinius ir mašininio mokymosi metodus /","Analysis of motor vehicle insurance premiums and claims risk using statistical and machine learning methods."],"dc:type":["info:eu-repo/semantics/bachelorThesis"]},"updated_at":"2026-11-19T10:12:55Z"}