Machine Learning Methods for Detecting Car Insurance Fraud: Comparative Analysis

Due to the massive premium payments and millions of annual claims, the automotive insurance business has expanded into a multi-million dollar industry. A sizable portion of the expense of owning a car goes toward vehicle insurance. Due to the fact that the insurance sector is one of the least predictable and revolves heavily around risk thus it is heavily reliant on data science as it makes the work a lot easier. This study examines the efficiency and verifiability of the most popular machine learning (ML) algorithms like Random decision Forest, Support Vector Machine (SVM) algorithm, logistic regression, XGboost, Naives Bayes, Decision tree, k Nearest neighbors etc. for fraud detection. Using the aforementioned techniques, we calculated the efficiency, recall, precision F1 score, and AUC score.