基于结构化大数据深度挖掘的非寿险保险公司经营风险模型研究

项目名称： 基于结构化大数据深度挖掘的非寿险保险公司经营风险模型研究

项目编号： No.61502280

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 计算机科学学科

项目作者： 闫春

作者单位： 山东科技大学

项目金额： 22万元

中文摘要： 本项目针对非寿险业务结构化大数据，按照保险经营管理的环节流程，主要针对核保、准备金评估、核赔阶段的风险进行非寿险经营风险建模研究，深度挖掘数据信息，解决目前存在的非寿险保险公司中大量的信息不能被充分利用的问题，为经营决策者提供更有效的决策支持。首先，从风险-贡献的双重视角，将模糊技术应用到关联规则的挖掘中，综合利用关联规则与知识分类模型刻画非寿险客户分类管理，进行交叉销售建模。然后，提出异常的、缺失的赔款数据下的非寿险准备金评估方法，利用孤立点挖掘方法检验索赔数据离群值进而提出准备金估计改进方法，引入平滑技术改进准备金评估模型的线性预估量的结构，增加模型的灵活性和预测功能。最后，采用极限学习机等数据挖掘技术进行非寿险反欺诈检测建模。本研究将为非寿险保险公司规避经营风险提供新的理论基础和技术支持，具有十分重要的理论意义和实践价值。

中文关键词： 大数据处理；数据挖掘；风险建模

英文摘要： The project studies business risks modeling of non-life insurance companies during the stage of underwriting, reserve evaluation and claim adjusting based on structured data of the non-life insurance business and according to the process of insurance management. The main purpose is to further mine data information, solve the problem that at present a large number of non-life insurance company's information can not be make full use of and provide more effective decision support for business decision makers. First of all, from the dual perspective of risk and contribution, fuzzy technology is used in mining the association rules. Association rules and knowledge classification models are both used to illustrate the classification management of non-life insurance clients and cross-selling modeling is done. Then, the assessment method of non-life insurance reserve is presented in the condition of abnormal and missing data. Outlier mining methods are used to test outliers of claims data and put forward the improved method for reserve estimation. Smoothing techniques are introduced to improve the structure of linear evaluation of reserve evaluation models and enhance the flexibility of the model and the prediction function. At last, some data mining techniques such as extreme learning machine are adopted to establish non-life insurance anti-fraud detection model. This study will provide new theoretical basis and technical support for the non-life insurance companies to avoid operating risks. This work has the very important theory significance and practice value.

英文关键词： Big data processing;Data mining;Risk modeling