A Study On Historical Order Integration For Insurance Company

Data mining is a important technology implemented in Customer Relationship Management and Marketing Research. In industries that focusing on individual or small business customers like insurance and banking, data analysis on customer data plays an influential role in corporation strategy as well as marketing plan. However, it's hard to mine the data of massive individual customers or small business customers especially for those orders placed before the CRM system was applied.In order to improve forecast of business data, the Business Intelligence system need to define which orders belong to one exact customer. One customer could open several accounts with different addresses, phone numbers and even names. The first step of customer data processing is to pre-process order data to identify which orders/accounts belong to same customer.In this paper, we will use na(?)ve Bayesian classifier, Decision Tree and Neural Networks to do machine learning with order matching result, and evaluate the performance of these three classifiers. The paper will provide algorithm introduction, steps for model establishing, performance evaluation and result. The tools and models will be implemented to a dataset of an insurance company for real business application. The models discussed can be implemented to different industries with different applications, including Internet, Banking, Insurance, etc.