Manuscript received August 27, 2024; revised September 25, 2024; accepted October 27, 2024; published December 20, 2024.
Abstract—Generalized linear model is a commonly used method in the traditional cumulative claim prediction. With the advent of the era of big data, machine learning algorithm has achieved good results in this field, but it is still not very good in prediction accuracy, fitting effect and practical interpretation. In the era of big data, with the substantial increase in the number and dimension of data, the key to how to predict the pure premium amount more accurately lies in finding a model that is more suitable for the characteristics of data, and improving the prediction effect and accuracy of the model is a crucial issue. In order to solve the problem of low prediction accuracy of accumulated insurance claims, a kind of data equalization method is adopted to predict the balanced insurance data, and the XGBoost model with improved loss function is used to predict the insurance data after SOMTE equalization. The improved model is more suitable for the processed data, and finally it has better effect and more accurate prediction results than the original XGBoost model.
Keywords—XGBoost, Synthetic Minority Over-sampling Technique (SMOTE), Huber loss, cumulative claim amount
Cite: Yang Li and Zhao Jinyan, "Prediction of Accumulated Auto Insurance Claims Based on Improved XGBOOST Modeling,"
International Journal of Engineering and Technology, vol. 16, no. 4, pp. 250-254, 2024.
Copyright © 2024 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).
