Design And Implementation Of Business Intelligence Analysis System Of A Telecommunication Company

With the development of information society and the continuous improvement of enterprise business level,it is difficult for decision support systems to continue to provide effective support for managers,so the ensuing business intelligence analysis system has very important application value.It obtains effective information from a large amount of data,helps users to understand the business situation in all directions,and then makes wise and practical decisions to make the business more efficient and fast.At present,business intelligence analysis systems on the market generally focus on data extraction,conversion,and loading,that is,ETL(Extract-Transform-Load)processing or self-service report analysis,resulting in many tools involved in the development process,which cannot be managed uniformly.Meanwhile,there are some limitations of single tool in practical application,such as ETL tool kettle lack of monitoring operation and maintenance service,and do not support kerberos authentication,unable to access the big data cluster that opened this authentication,which inevitably increases the difficulty difficult for developers to work.In order to solve the above problems,this paper designs a business intelligence analysis system based on big data technology.In this system,I mainly participated in the design and development of the kettle ETL platform,data mining,reporting platform,data source management and authority management functions,and used the front and rear end separation technique to effectively decouple the interface display and business logic processing.Among them,the kettle ETL platform completes data conversion processing operations through drag and drop and component configuration,and implements kerberos certification support,task scheduling,and monitoring management functions.The data mining module mainly based on the subject of customer loss in the data warehouse,using random forest,XGBoost(e Xtreme Gradient boosting)and other algorithms to realize the construction of churn prediction model,and using K-means algorithm to subdivide customers,combined with churn factor analysis and group characteristics to give the corresponding customer care for potential churn customers.The report platform is mainly responsible for the visualization process of data,including data modeling,multidimensional analysis and data report customization.Authority management provides the security of data,which is mainly responsible for the allocation and control of system resources,including three modules: role management,menu bar management and authority control.The research and development process of the system is mainly based on the Spring Boot framework,using My SQL and HDFS(Hadoop Distributed File System)as data storage tools,and use tools such as Apache Kylin to complete the data calculation.Finally,the micro service architecture is used to integrate various functions into one to provide users One-stop application service improves the efficiency of statistical analysis.At present,the system has been initially on line,running in good condition.At the bottom of the system is connected to the company's big data platform,and the data processing and analysis are done under the platform,so the efficiency of the system is about 10 times higher than before.At the same time,the system will integrate kettle ETL? data mining and visual analysis seamlessly,and then can complete the data processing and analysis display process more conveniently.