Insurance Claim Fraud Detection using Benford’s Method and Machine Learning

Insurance claim fraud poses significant challenges for insurance companies, leading to significant financial losses, reputation damage and declining the trust of policyholders. Detecting fraudulent claims is of paramount requirement in the financial service sector, which also requires advanced analytical techniques and sophisticated fraud detection models. This research paper aims to investigate and analyze insurance claims fraudulent patterns using a comprehensive dataset comprising historical claim records and associated attributes. Benford’s method is used for identifying fraudulent insurance claims along with statistical analysis using machine learning algorithms. The proposed methodology promises an accuracy of 98.47% by random forest, 83.31% in linear regression and 95.21%, by using an artificial neural network, which indicates the model accurately identifies the fraudulent insurance claim.