Interpretable Property Insurance Premium Prediction Using Machine Learning Models

Accurately predicting home insurance premiums is a critical challenge for insurers, as traditional methods struggle with the complexity and volume of modern data. This study leverages machine learning to address this problem, applying a range of supervised models—including linear regression, lasso regression, ridge regression, decision trees, random forests, gradient boosting, and extreme gradient boosting (XGBoost)—to a comprehensive home insurance dataset with 66 features. After preprocessing to retain 50 key variables, the models were trained and evaluated, with random forests, gradient boosting, and XGBoost emerging as top performers, achieving R-squared scores of 0.8137, 0.8014, and 0.8344, respectively. Hyperparameter tuning further improved XGBoost’s performance to an R-squared of 0.8380,