Modeling the Frequency and Severity of Auto Insurance Claims using Machine Learning Techniques

The auto insurance industry constantly seeks innovative approahes to accurately predict and manage both the frequency and severity of insurance claims. This study delves into leveraging machine learning techniques to model insurance claims, focusing on estimating the frequency and severity of claims. A comprehensive dataset encompassing a wide array of variables related to policyholders, vehicles, accidents, and historical claims was used to train and validate the machine learning models. For a frequency prediction, models such as poisson regression, decision tress, random forests, and gradient boosting were employed to estimate the likelihood of a claim occurence. The severity prediction entailed the application of regression-based models, such as linear regression and decision trees, for the purpose of forecasting the financial magnitude of claims in the event of their occurrence. The results demonstrate the efficacy of machine learning in accurately predicting the frequency and severity of auto insurance claims. The predictive models achieved notable accuracy and performance metrics, aiding insurers in assesing risk, setting premiums, and optimizing their claim handling processes. Moreover, this study illuminates the prospect of augmenting decision-making and risk management capabilities in the automobile insurance sector by incorporating advanced machine learning methodologies. Mathematics Subject Classification: 34A08, 65M06, 65N12, 35R11.