Research On The Key Technology Of BI System Based On Big Data Technology

With the rapid popularization and penetration of the Internet and mobile Internet, Global data grows explosively, the era of big data has arrived. How to use the vast amounts of data to promote the development of themselves has become a major problem for the companies. In the past, companies usually analyse their data shallowly by the BI system with a small amount of data. BI is short for Business Intelligence, it is a complete solution for companies to effectively integrate their existing data, providing fast and accurate report forms and recommented decisions to help companies to make wise business decisions. Traditional BI system mainly analyse the data by report form, data warehouse and data mining. However, in the era of big data, BI system’s analysis and processing capabilities appeared huge bottleneck due to the rapid increasing amount of data. In the context of big data era, it’s urgent to upgrade the BI system throught the storage technique, handing system and analysis technique, in order to adapt the trend of technology’s development. However, the research of BI system based on big data’s technique is rare at present.For the needs of integrating BI system and big data technology, this paper design a BI system based on big data technology named “BigData-BI”. The system implements the functions included data acquisition, data processing and data analysis, using big data’s related technologies and architectures to improve the traditional BI system. The main contribution od this paper is using the “BigData-BI” system to analyse three main business of China Life Insurance Guangdong Branch, providing powerful help for the company to make sales strategy, including: Firstly, specific to the situation of insurance salesman saling products to customers blindly, “Genealogy mining algorithm by GraphX” is designed to mine customers’ genealogy, making the insurance saleman can clearly know the family situation of the customers in order to sell insurance products more precisely to customers and their families. Secondly, specific to the extensive screening target customers’ method, “Random Forest algorithm based on slicing technique” is designed to scientific classification of customers, helping China Life Insurance Guangdong Branch screening high-quality leads and greatly enhancing the efficiency and accuracy of sales. Thirdly, specific to the lacking test mechanisms for the past sales strategies, “FP-Growth association rules mining algorithm based on memory computing” is designed for the sales strategies regression test, helping China Life Insurance Guangdong Branch to analyse the impact for the sales’ behavior by every features in the strategies and providing powerful reference for the next quarter’s sales strategies.Currently, the system is set up and experimental running in China Life Insurance Guangdong Branch. The analytic result of three algorithms has been used to the next quarter’s strategies making. System is running smoothly and effectively, fully meeting the expectations.