A novel data processing approach to detect fraudulent insurance claims for physical damage to cars

Some automobile insurance companies use computerized auto-detection systems to expedite claims payment decisions for insured vehicles. Claims suspected of fraud are evaluated using empirical data from previously investigated claims. The main objective this manuscript is demonstrate a novel processing system and its potential in classification. approach was used develop machine learning-based sentiment classification model describe property damage vehicle accidents the indicators fraudulent To end, Singular Value Decomposition-based components correlation-based composite variables were created. Machine learning models then developed, with predictors selected based on standard feature selection procedures. Five used: Boosted Trees, Classification Regression Random Forests, Artificial Neural Networks, Support Vector Machines. For all models, achieved higher accuracy rates, among these artificial neural network highest performance at 76.56%.