智能车险定价车辆信息数据

通过公司历年数据中对车辆从人信息、从车信息、从投保险别信息、从历史承保信息、从平台提供信息进行清洗归集,建立广义线性模型建模，量化费率因子的系数水平，计算各类保单的真实成本和建议费率从而实现精细化、差异化定价。通过对不同场景的深入分析，在细分领域为中小型保险公司提供可参考的定价方式和定价结果，促使中小型保险公司提供更加贴近客户需求和风险状况的产品，更有效地配置资源，实现车险精细化经营。对于保险用户而言，风险较低的优质客户可对应更低的保费，享受实实在在的优惠，对于风险较高的客户适当提高价格对后续在从事交通活动中起到警示作用。1.数据采集：清洗出本公司历年承保车辆投保、承保、车辆信息以及相应平台信息作为定价模型的x变量(具体可见其他说明)。
2.数据处理: 分险种代码应用广义线性模型，广义线性模型的基本形式为 g(y_hat) = ax + b，其中 g() 是连接函数，y_hat 是目标变量赔付成本。上述变量中车型库相应值为空的情况下取核心录入数据，如座位数值获取：【座位数（车型库）】值为空时则取【座位数（核心）】。a 和 b 是模型参数。连接函数 g() 根据数据车险定价分布一般假设服从tweedie分布，连接函数选择对数函数，通过广义线性模型极大似然估计返回上述x变量对应的系数。通过乘法定价公式(将上述x变量对应系数进行相乘)得到[赔付成本](中间变量)；[无赔款优待系数](中间变量)由【无赔款优待及上年赔款记录】根据码表转换。[成本系数](中间变量)=[赔付成本]/公司设定目标赔付率(固定值) -【签单保费】/(【标准保费】*[无赔款优待系数])，结合[成本系数]【业务性质】、【投保人客户标识】通过码表转换得到最终自主定价系数。
3.数据应用：新投保车辆通过相关信息匹配相应车险因素得出相应的自主定价

By cleaning and aggregating person-related vehicle information, vehicle-related information, insurance coverage information, historical underwriting information, and platform-provided data from the company's annual historical datasets, we establish a Generalized Linear Model (GLM) to quantify the coefficient levels of rate factors, and calculate the true cost and recommended premium for each insurance policy, thereby achieving refined and differentiated pricing. Through in-depth analysis of different scenarios, we provide reference pricing methods and results for small and medium-sized insurance companies in segmented fields, enabling these companies to launch products that better align with customer needs and risk profiles, allocate resources more efficiently, and realize refined auto insurance operations. For insurance policyholders, high-quality low-risk customers can enjoy lower premiums and tangible discounts, while appropriately raising prices for high-risk customers will serve as a deterrent for their future traffic-related activities.

1. Data Collection: Clean and aggregate the policy application, underwriting, and vehicle information of the company's over-years' insured vehicles, as well as corresponding platform data, as the X variables for the pricing model (details refer to other supplementary instructions).
2. Data Processing: Apply the Generalized Linear Model by insurance type code. The basic form of the Generalized Linear Model is $g(hat{y}) = ax + b$, where $g()$ is the link function, and $hat{y}$ is the target variable claim cost. If the corresponding value in the vehicle database is missing, use the core entry data instead. For example, when the value of [Number of Seats (Vehicle Database)] is empty, use [Number of Seats (Core Entry)]. $a$ and $b$ are model parameters. The link function $g()$ is set as the logarithmic function, assuming that the auto insurance pricing distribution follows the Tweedie distribution. The coefficients corresponding to the aforementioned X variables are obtained via maximum likelihood estimation of the Generalized Linear Model. The [claim cost] (intermediate variable) is calculated through the multiplicative pricing formula (multiplying the coefficients corresponding to the above X variables); the [no-claim discount (NCD) coefficient] (intermediate variable) is converted from the [no-claim discount and previous year's claim records] based on the official code table. The [cost coefficient] (intermediate variable) is calculated as: $ ext{[cost coefficient]} = ext{[claim cost]} / ext{(company-set target claim ratio, a fixed value)} - ext{[written premium]} / ext{([standard premium]} * ext{[no-claim discount coefficient])}$. The final autonomous pricing coefficient is obtained by converting the [cost coefficient], [business nature], and [policyholder ID] through the code table.
3. Data Application: For newly insured vehicles, match the corresponding auto insurance factors based on their relevant information to derive the corresponding autonomous pricing.

搜集汇总

数据集介绍

特点

该数据集是用于智能车险定价的车辆信息数据，包含3001条记录，每月更新，涵盖车辆信息、保险信息等多个维度，通过广义线性模型建模实现精细化定价。