【作者】 张潇；

【导师】 张茜；

【作者基本信息】 山东大学 ， 保险专硕（专业学位）， 2020， 硕士

【副题名】基于随机森林模型

【摘要】 自改革开放以来,我国经济迅速增长,良好的经济环境带动各行各业的蓬勃发展,与此同时我国汽车保有量也随之大幅度提高。2001年12月,我国正式加入W TO世界贸易组织,汽车保险作为我国最重要的险种之一,首先实行对外资保险公司开放。由此可见,汽车保险对于我国财产保险甚至整个保险界,占有重要的战略地位。自2015年6月起,我国银保监会启动新一轮商车费改,截止到2018年3月8日,银行保险监管部门先后实施了 3次商车费改,其发布文件中规定,财产保险公司计算商业汽车保险保费时在费率厘定方面可以自主进行确定费率因子。但与此同时,因为定价范围的扩人市场竞争更加激烈,许多公司为抢占市场打价格战,全国几乎所有中小型财产保险公司年利润额为负的状态,这对我国保险业的发展极为不利。根据现实情况,对容户建立一套有效的车险索赔概率预测模型,无论是整个保险行业还是各大财产保险公司都是一件当务之急的大事。随着数据时代的来临,大数据已经为各行各业的发展带来了转变的契机。因此,如何利用大数据技术制定更为精准的汽车索赔概率模型已成为各家保险公司发展的关键。本文结合时代背景,运川定性分析法、比较分析法、文献研究法对于大数据背景下如何更为精准高效预测车险索赔概率问题进行研究。论文首先详细介绍了我国机动车辆保险市场发展状况以及国外主要保险市场的车险费率改革方法,并且简单介绍如今国外基于大数据技术所研发的保险产品现状。之后,对传统Lo gistie模型以及随机森林模型进行详细阐述,分析其模型间的优劣性。然后在大数据背景下,不仅预测精确度需要考虑,计算效率也需要同样重视,通过Gini值法进行变量筛选可以解决此问题。因此本文利用实际数据进行的实证检验部分中,主要包括了两个部分,一是验证了随机森林模型在车险索赔概率预测问题上具有良好的精确度,二是进行变量筛选后,随机森林模型同样具有较高的效率。本文的研究结果可以应用与保险行业中的汽车索赔概率的预测,为客户个性化化定价提供理论依据,通过对低风险的客户降低保费,对高风险的客户提高保费,从而吸引更多优质客户。于此同时,本文所建立的模型可以为保险领域的其他预测问题研究提供一定参考。最后,本文对全文进行概括总结,并针对国内保险界现状提出个人建议。更多还原

【Abstract】 Since 2000,with the rapid development of China’s economy and the increase of per capita disposable income,the per capita ownership of automobiles has grown rapidly.At the same time,due to the support of relevant national policies,the entire automobile insurance industry has developed comprehensively.With the rapid economic development and a series of policy support,the domestic auto insurance industry has developed rapidly,and auto insurance has become the largest insurance type of property insurance premium income,so it has also become an important part of daily life.Since the start of a new round of commercial fare reform in June 2015,as of March 8.201 8,the insurance regulatory department has implemented three commercial fare reforms,and has adopted a gradual approach to pilot commercial fare reforms in various provinces.Allowing insurance companies to determine the scope of the adjustment of the rate coefficient themselves when drafting commercial vehicle insurance rates.However,at the same time,the commercial car fee reform has also intensified competition among various insurance companies.Most insurance companies blindly rely on low premiums to obtain customers and high commissions to expand intermediary channels,and ignore the potential risks of customers and the company’s own operations Risks,leading to the loss of annual profits of most small insurance companies.Therefore,insurance companies urgently need a set of auto insurance claim probability prediction models that can calculate more scientific and reasonable auto insurance premiums based on the risk levels of different customers.With the advent of the data age,big data has brought opportunities for change in the development of various industries.Therefore,how to use big data technology to develop a more accurate probability model of automobile claims has become the key to the development of various insurance companies.The content of this article involves theoretical knowledge in insurance.statistics,actuarial science,and computers.The qualitative analysis method,comparative analysis method,and literature research method are used to study the application of machine learning models to the prediction of automobile insurance claims.The dissertation first introduces the development status of the motor vehicle insurance market in detail.Nowadays,the development of big data insurance at home and abroad,these basic conditions can be verified by understanding the examples later.Secondly.this article takes an insurance company in Shandong as an example to empirically analyze the application of big data technology to insurance companies.The empirical results show that the random forest algorithm has a good performance in the auto insurance claim probability prediction model.The research results in this paper can be applied to the prediction of vehicle risk in the insurance industry,providing a basis for differentiated pricing for customers,reducing premiums for low-risk customers,and increasing premiums for high-risk customers,thereby attracting more high-quality customers.At the same time,the research method in this paper can provide a reference for other forecasting problems in the insurance field.Finally,this article summarizes the full text,and puts forward personal recommendations for the current status of domestic insurance.更多还原