A Machine Learning and Explainable Artificial Intelligence Approach for Insurance Fraud Classification

This study tackles the pressing issue of fraud in the vehicle insurance market by introducing a comprehensive framework that integrates advanced detection models with Explainable Artificial Intelligence (XAI) and heterogeneous classifiers. The inclusion of XAI is particularly significant, as it enhances the interpretability and transparency of machine learning algorithms, which is crucial for maintaining the integrity and trustworthi-ness of insurance operations. Our methodology employs three distinct XAI techniques: Shapeley Additive Values (SHAP), Explain Like I’m 5 (ELI5), QLattice and Local Interpretable Model-Agnostic Explanations (LIME) to elucidate the decision-making process of machine learning models. This approach not only ensures model inter-pretability but also identifies key factors influencing fraud detection, including vehicle age, base policy, fault, deductible, and policyholder age. The standout contribution of our research is the development and validation of a multi-stack machine learning model that achieves an exceptional accuracy rate of 96 %, significantly outper-forming traditional classifiers. This high level of accuracy, combined with the interpretability provided by XAI, underscores the potential of our framework to revolutionize fraud detection practices in the vehicle insurance sec-tor. By offering a robust, accurate, and interpretable solution to fraud detection, this study makes a meaningful contribution to the field. It provides valuable insights and tools for insurance providers aiming to enhance their fraud detection capabilities, setting a new benchmark for the development of advanced, reliable systems in the industry.