多维分析系统——MDA4YZD的设计与实现-计算机技术专业论文.docxVIP

万方数据 万方数据 关键词：数据仓库；数据清洗；多维分析；缓慢变化维IV 关键词：数据仓库；数据清洗；多维分析；缓慢变化维 IV Abstract The Data Warehouse and OLAP (On-Line Analytical Processing) technologies have gained continual development and advancement in recent years. At present，plenty of govemments and enterprises have already established their own data warehouses，providing their own development and leading decisions with significant data support Currently，there has been a mature and stable business OLTP(On-Line Transaction Processing) system in domestic supervising field of precursor chemicals，which meets the needs from the clients requirements ，helps the online approval for the supervising departments and improves the efficiency of the supervisors. However，it is impossible for the OLTP system to provide analysis for a large amount of historical data since it can simply satisfy common and daily operations. Aiming at this problem ，Hebei SiBosi Innovation Technology Co.，Ltd. decided to separate OLTP and OLAP system ，adapting data warehouse and multi-dimensional analytical processing and develop a multi-dimensional analysis system(MDA4YZD).τhe main contents are listed as follows: Establishment of a data warehouse with sales and storage orientation for the core supervising business of precursor chemicals. In order to reduce space waste and improve query efficiency，the Snowstorm schema is adopted ，as well as a multi-dimensional model which accords with the ROLAP-based field background as the saving mode is initiated. As for those unforeseen dimensional multi-variation problems，a solution based on timestamp is proposed. Development of an ETL subsystem for extracting，c1eaning，transforming and loading data according to the original data features ofthe OLTP system. To solve the redundant noise data problems including enterprises info and etc，hierarchical filters on the basis of business principles for providing high quality data are designed. Development of a multi-dimensional analysis visualization subsystem applying Bootstr