Design And Implementation Of A Reports System Based On Data Mining

Recently, with the rapid developing of data house and data mining and consideration of cost and data security, more and more companies, especially IT companies and software companies start to build their own data-driven operation platform. Reports system is on the top tier of the tree-tire data warehouse architecture. At the same time, it is also one of the most important sub-systems of data-driven operating platform. This paper design and implementation a reports system based on the data-driven operating platform which owned by the wire-less operation department of a big Internet company.Firstly, after analyzed the existing data-driven operation system which based on data warehouse, this paper discussed the reports system’s user requirements and cleared the function boundary of the reports system by modeling method of software engineering. It also described a lot about the requirement of data prediction function of the reports system which based on data mining method. What’s more, it modeled the report configuration object.Secondly, this paper designed the system based on requirements analysis. It designed the architecture of the reports system based on the three-tier MVC model, discussed an improved two phrase report generation method and utilized object-oriented method for analyzing and designing the reports configuration function, the reports display function, the data prediction function and system management functions. The reports configuration function designing included configuration saving and loading, real time previewing and visual layout function. The reports display function designing included table generating, chart generating, table exporting and chart exporting. Data prediction function designing described the data prediction function which was integrated in the reports system for time series prediction based on Support Vector Machine.In the implementation procedure of the system, this paper not only built the application on JavaEE platform, used HighCharts programming library to display charts in the reports system, implemented data prediction function which was base on SVM with LIBSVM tools, applied some actual index and operation data of products to verify the data prediction function which based on SVM and integrated in the reports system.Lastly, this paper tested the functions implementation and the response time of the system. According to the testing results, this paper proved that the implementation of the reports system has achieved the requirements which were discussed in requirements analysis. After implementation, this system was adopted by one of the famous Chinese internet company as daily operation reports system. According to user’s feedback, the system supported the data analysis jobs of users. Therefore, it proved that the solution provided by this paper is effective.