当前行业测算的新能源车险基准保费属于车型定价范畴，考虑风险因子不全面，且测算所用广义线性模型仅能识别纯风险损失与定价因子间的线性关系，准确性有待提高。对此，本文在车型定价基础上，将“三电”系统参数、驾驶行为因子、充放电习惯因子、周围环境因子纳入定价因子范围，并使用随机森林算法建立新能源车险定价模型，以提高定价的公平性和科学性。结果表明，品牌、险种决定保费的基准水平，里程、夜间行驶、速度等驾驶行为因子是影响保费波动的重要因素，充电行为因子和周围环境因子是影响保费波动的次重要因素。同时，比起GLM，随机森林有更好的拟合效果和风险区分能力，且随定价因子变化，预测保费与真实损失呈现相同趋势。

The base premium of new energy vehicle insurance which is widely used by insurance companies was rated considering brand only and using GLM which can identify the linear relationship between loss and pricing factor only. To improve the fairness of premium, this thesis consider pricing factors about environment, electricity, drivers, charging and discharging habits, and construct a new energy vehicle insurance pricing model based on random forest. The results show that brand and type of insurance determine the level of premium, factors about drivers such as mileage, night driving, and speed are important factors which affect premium, and factors about charging habits are secondary important. Moreover, compared with GLM, random forest fits better and has larger premium discrimination between clients of low and high risk, and the prediction shows the same trend as the loss as a factor changes.