A Comparative Study of Machine Learning Algorithms for Vehicle Insurance Fraud Detection

Manthena, Shruthi (2024) A Comparative Study of Machine Learning Algorithms for Vehicle Insurance Fraud Detection. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (2MB) | Preview

Abstract

Vehicle insurance fraud detection is the biggest problem for insurance organizations, because scammers cause great losses and insurance is becoming more expensive for loyal customers. Traditional methods cannot scale well, read poorly and cannot adequately address large volumes of complex data. This study explores the application of machine learning models for predicting insurance claims and fraud detection across three distinct datasets: Claim Fraud Identification within the Insurance Industry, Automobile Insurance Information and Automobile Insurance Claim Forecasting. To get started data cleaning and preprocessing techniques are applied, which involve prominent missing values, encoding critical categorical features and scaling the features if required. Model selection methodologies were used to improve feature selection. Five machine learning algorithms namely Decision Tree, K- Nearest Neighbors (KNN), Light Gradient Boosting Machine (Light GBM), Random Forest and Support Vector Classifier (SVC) were used with each dataset to classify insurance outcomes. To evaluate model efficiency, accuracy together with F1 score, precision and recall were calculated. For Insurance Fraud Claim Detection set, it was determined that Light GBM has the best performance with F1 score of 0.612 and a precision of 0.6. In Car Insurance Data case, Light GBM was the best with an accuracy of 0.841 and F1 score of 0.744. Last but not least, the Light GBM model showed 100% accuracy, F1 score, precision and recall in the vehicle insurance claim prediction dataset to become a model of choice once more across the three simulations. The findings presented in this study clearly illustrate the prospects of machine learning algorithms, especially Light GBM for improving the effectiveness of insurance claim prediction and fraud detection.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Menghwar, Teerath Kumar UNSPECIFIED
Uncontrolled Keywords:	Vehicle Insurance Fraud Detection; Machine Learning; Fraudulent Claims; Insurance Data; SVC (Support Vector Classifier)
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science H Social Sciences > HG Finance > Insurance Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning H Social Sciences > HD Industries. Land use. Labor > Specific Industries > Motor Industry
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	03 Sep 2025 13:42
Last Modified:	03 Sep 2025 13:42
URI:	https://norma.ncirl.ie/id/eprint/8745

Actions (login required)

View Item