Strategies for Predicting and Managing Life Credit Insurance Claims Using Machine Learning Models

Abstract

The rising severity of credit life insurance claims makes it necessary to develop new methods for managingclaims effectively. Machine Learning (ML) offers a powerful response to this challenge. Since improving customer service is a central objective for insurers, companies have increasingly turned to ML techniques to better understand and assess their data. This paper makes a scientific contribution to life credit insurance pricing by modelling the total claim amount using ML algorithms such as Support Vector Regression (SVR), Extreme Gradient Boosting (XGBoost) and Multi-Layer Perceptron (MLP). A comparative analysis is carried out using statistical performance measures (e.g., Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), etc.) on both training and testing datasets. The results indicate that the Multi-Layer perceptron model delivers excellent predictive accuracy and outperforms the other models according to the Taylor diagram. However, its visual distribution of predictions is less satisfactory than that of the XGBoost and MLP models. Overall, the main value of this study lies in the in-depth analysis of the dataset, which provides insurers with meaningful insights to support more effective loss management.