Artificial intelligence for fraud detection in motor insurance sector

Detalhes bibliográficos
Autor(a) principal: Simão, Carolina Gonçalves Silva
Data de Publicação: 2022
Tipo de documento: Dissertação
Idioma: eng
Título da fonte: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
Texto Completo: http://hdl.handle.net/10362/133845
Resumo: Project Work presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Data Science
id RCAP_a79f251a9f43322895ee38d698383c88
oai_identifier_str oai:run.unl.pt:10362/133845
network_acronym_str RCAP
network_name_str Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
repository_id_str https://opendoar.ac.uk/repository/7160
spelling Artificial intelligence for fraud detection in motor insurance sectorFraudInsuranceAutomobileMachine LearningPythonProject Work presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Data ScienceOne of the major problems in the insurance sector is related to fraud, aside from tax fraud, insurance fraud is the most practiced fraud in the world since insurance, by its nature is very susceptible to it. Fraud could be minimized by investigating each claim that occurs but that also means an increase of the costs for the insurance companies. The fraudulent clients or agents that will be caught with the investigation and the amount of money spent by looking into every new claim is not worth it. Insurance fraud is usually caught only when the fraudsters get greedy and it becomes obvious that they are involved in a scheme. To minimize the investigation costs by only looking at suspicious claims, this project tries to identify the ones that are worth to scrutinize, through machine learning techniques. Five different predictive models will be used: Logistic Regression, Decision Tree, Random Forest, Neural Network and Gradient Boosting. The goal is to build an optimal model that will determine which automobile claims have higher probability of being fraudulent. An efficient fraud management can reduce costs, minimize claims and increase profits. This goal was accomplished with a Gradient Boosting classifier with 400 estimators, that is able to predict correctly 49% of the fraudulent claims, with 75% less investigated claims. There is still room for improvement by introducing the expected claim and investigation costs in the model. Since only the ones with significant costs would be worth to open an investigation, an even greater decrease in the number of investigated claims would be possible and, consequently, a decrease in the company’s costs with claims. Also, it would be expected that the claims with higher costs are more likely fraudulent than the ones with small indemnities; hence, this variable could lead to a higher precision of the model. These two features will be available in the future.Gonçalves, Rui Alexandre HenriquesRUNSimão, Carolina Gonçalves Silva2022-03-03T17:21:38Z2022-01-282022-01-28T00:00:00Zinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesisapplication/pdfhttp://hdl.handle.net/10362/133845TID:202966640enginfo:eu-repo/semantics/openAccessreponame:Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)instname:FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologiainstacron:RCAAP2024-05-22T17:59:56Zoai:run.unl.pt:10362/133845Portal AgregadorONGhttps://www.rcaap.pt/oai/openaireinfo@rcaap.ptopendoar:https://opendoar.ac.uk/repository/71602025-05-28T17:30:52.450503Repositórios Científicos de Acesso Aberto de Portugal (RCAAP) - FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologiafalse
dc.title.none.fl_str_mv Artificial intelligence for fraud detection in motor insurance sector
title Artificial intelligence for fraud detection in motor insurance sector
spellingShingle Artificial intelligence for fraud detection in motor insurance sector
Simão, Carolina Gonçalves Silva
Fraud
Insurance
Automobile
Machine Learning
Python
title_short Artificial intelligence for fraud detection in motor insurance sector
title_full Artificial intelligence for fraud detection in motor insurance sector
title_fullStr Artificial intelligence for fraud detection in motor insurance sector
title_full_unstemmed Artificial intelligence for fraud detection in motor insurance sector
title_sort Artificial intelligence for fraud detection in motor insurance sector
author Simão, Carolina Gonçalves Silva
author_facet Simão, Carolina Gonçalves Silva
author_role author
dc.contributor.none.fl_str_mv Gonçalves, Rui Alexandre Henriques
RUN
dc.contributor.author.fl_str_mv Simão, Carolina Gonçalves Silva
dc.subject.por.fl_str_mv Fraud
Insurance
Automobile
Machine Learning
Python
topic Fraud
Insurance
Automobile
Machine Learning
Python
description Project Work presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Data Science
publishDate 2022
dc.date.none.fl_str_mv 2022-03-03T17:21:38Z
2022-01-28
2022-01-28T00:00:00Z
dc.type.status.fl_str_mv info:eu-repo/semantics/publishedVersion
dc.type.driver.fl_str_mv info:eu-repo/semantics/masterThesis
format masterThesis
status_str publishedVersion
dc.identifier.uri.fl_str_mv http://hdl.handle.net/10362/133845
TID:202966640
url http://hdl.handle.net/10362/133845
identifier_str_mv TID:202966640
dc.language.iso.fl_str_mv eng
language eng
dc.rights.driver.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
dc.source.none.fl_str_mv reponame:Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
instname:FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologia
instacron:RCAAP
instname_str FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologia
instacron_str RCAAP
institution RCAAP
reponame_str Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
collection Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
repository.name.fl_str_mv Repositórios Científicos de Acesso Aberto de Portugal (RCAAP) - FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologia
repository.mail.fl_str_mv info@rcaap.pt
_version_ 1833596748438700032

Registros relacionados