大力发展新能源汽车是我国推动绿色发展，建设美丽中国的重要举措。近年来新能源汽车的保有量快速增加，但在新能源车险定价中存在定价因子多，定价模型需要改进的问题。本文创新地提出了一个双层模型对新能源车险进行定价，该双层模型基于广义线性模型和XGBoost算法。使用广义线性模型作为底层输出可以保证模型的可解释性，加入XGBoost算法之后又可以增加模型的预测精度，因此双层预测模型具有良好的性能。基于一组实际的新能源汽车理赔数据进行了实证分析，实证结果表明双层预测模型的预测效果更好，贝叶斯信息准则和模型偏差度均小于单一的广义线性模型，并且双层模型能更好地预测极端损失。

Vigorously developing new-energy vehicles is an important measure for our country to promote green development and build a beautiful China. In recent years, the number of new energy vehicles has increased rapidly, but there are many pricing factors in the pricing of new energy vehicle insurance, and the pricing model needs to be improved. This paper innovatively proposes a two-layer model for pricing new energy vehicle insurance based on generalized linear model and XGBoost algorithm. The generalized linear model can be used as the bottom output to ensure the interpretability of the model, and the prediction accuracy of the model can be increased after XGBoost algorithm is added. Therefore, the two-layer prediction model has good performance. Based on a group of actual claims data of new energy vehicles, empirical analysis is carried out. The empirical results show that the prediction effect of the two-layer prediction model is better, the Bayesian information criterion and the model deviation are smaller than the single generalized linear model, and the two-layer model can better predict the extreme loss.