基于数据优化的保险客户承保预测

Insurance Customer Purchase Prediction Based on Data Optimization

近年来,人民生活水平的普遍提高,使得保险行业迎来了新的春天。一直以来的粗放式经营模式已经无法满足保险公司日益发展的要求。如何摆脱传统的营销方式,快速发掘出有价值的客户,在市场中不远远落后,对于保险公司来说越来越重要。本文使用某人寿保险公司的客户数据。首先,基于给定的客户基本信息、通话信息、投保信息、赠险信息等进行描述性统计分析,查看数据情况,对数据进行数据清洗提升数据质量;其次,使用单独的逻辑回归模型进行学习,生成可行性分析报告;然后,分别使用决策树与逻辑回归的组合模型以及随机森林与逻辑回归的组合模型进行预测;最后,将三种模型进行对比发现随机森林与逻辑回归的组合模型效果更好。

In recent years, with the general improvement of people's living standards, the insurance industry ushered in a new spring. The extensive business model has been unable to meet the requirements of the increasing development of insurance companies. How to get rid of the traditional way of mar-keting, quickly discover valuable customers and keep up with the market, is becoming more and more important for insurance companies. This article uses customer data from a life insurance company. Firstly, descriptive statistical analysis was conducted based on the given basic information of customers, call information, insurance information and risk donation information, etc., to view the data situation, and data cleaning was carried out to improve the data quality. Secondly, a separate logistic regression model is used for learning to generate a feasibility analysis report. Then, the combined model of decision tree and logistic regression and the combined model of random forest and logistic regression were respectively used for prediction. Finally, a comparison of the three models shows that the combined model of random forest and logistic regression is more effective.