Based On Data Mining Technology, Insurance Renewal Reminders Method

Insurance has became a fast-growing business in China, with its increasing, a huge amount of customer information has been gathered together.The speed of the company' s intranet become higher and higher,the capability of the business-process server become more powerful, data-centralization become more popular.These provide a hardware platform for the application of data-mining technique in insurance business. More and more insurance company has realized the importance of the application of data-mining, both in risk-controlling and cost-reducing.With the development of insurance , the running expenses of insurance company is growing fast. The profit of insurance company has two part,one is the First-year Premium(FYP),the other is the renew premium. The renew premium has the property of stability and continuity, it has been a strong foundation of the insurance company to keep developing. Nowadays the way of getting renew premium in insurance company is very simple , for every customer,they give notice in the same way, it cost a lot and waste a lot.This paper try to use data mining technique and classification method to analyze the activities of customer in payment, to discover the relationship between the possibility of payment and how many times they have paid and when they have paid the premium. According to the information of customer , we partition the customer in several group, such as high quality customer, quasi high quality customer, faithful customer and risk customer. Using decision tree and multi-variable regression , we make a prediction model to estimate the possibility of a certain customer' s activity in payment. When the model is estabilished , it can give some advice to the renew department to adopt flexible method in getting renew premium. It make the cost more efficiency and may get more marginal profit.At the same time , we try to use clustering technique to identify the high-risk surrender customer from the large amount of customers, then insurance company will afford some special service for thosecustomers in order to reduce the rate of surrender.According to the model of this paper,we apply it in SMS system, then we can validate the model in this system. And we will keep improve the model based on the fact.