Font Size: a A A

Algorithm Based On Association Rules In Data Mining Research And Application

Posted on:2010-06-29Degree:MasterType:Thesis
Country:ChinaCandidate:J Q YuanFull Text:PDF
GTID:2178360278960509Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
By the end of the twentieth century, traditional database management systems had been in depression, the brutal competition in the market forced the world's supplier of database technology who had to search for other new economic growth point. One of which is well-known hotspot database vendors, such as Oracle, Sybase and other large database for a typical representative of the object - relational database technology .And the other hot spots in data warehouse technology is at the core Business Intelligence (Business Intelligence, BI).On-line Analytical Processing (OLAP) and data mining technology are increasing going to mature. The application for the data warehouse market development has laid a good foundation. Enter the twenty-first century, The on-line analytical processing and the data mining applications ,which based on the warehouse,in different areas in the country have managed to be applied to a wide range of application, such as finance, insurance, securities, telecommunications, tobacco, tax and other traditional data-intensive industry larger development. However, to compare with the developed countries ,there is relatively less than in both the basis of current information technology equipment and some ideas. To face the ever-growing historical database, data sources and alternative systems can not fully meet the requirements of flexibility .As a result ,that when the existing data warehouse based on the demand for new value-added services or when the algorithm the sustainable development of the whole system become the bottleneck.Users, mining database with the aid of the data analysis tools which are provided by database manufacturers, were found that both the consumption of time and the found useful business rules are not full satisfied. This thesis describes a large-scale analysis of supermarket chains operating system, data warehouse development and data mining research, analysis based on Boolean association rules mining algorithm Apriori exists a large chain of various inadequacies. The corresponding optimization algorithm is the Algorithm FP-Tree based on the FP-T solution. Combing with the ETL (Extract, Transform, Cleansing, Load) rules, we have managed to extract the extraction's strategy. At same time, we have discussed the data warehouse system design based on the realization of three-particle size, including detailed data on monthly aggregated data, aggregate data size year-on-year options, and how reasonable to achieve the dimension table and fact table granularity of the division of this issue of the composite primary key table longitudinal and cross-conversion between the table in order to achieve the dimension table and fact table size down the idea of a breakdown. That is, without modifying the source data systems, to maximize use of existing data to achieve the new requirements. The system development platform is in Windows operating systems, the use of Microsoft SQL Server for database storage, and use Microsoft SQL Server provides the SQL Analysis Processing Modeling Tools for Data Warehouse and ETL to achieve its operation, the use of Microsoft Office 2003 to achieve its front-end display tools. The entire design of data warehouse system and the research have been made an in-depth analysis and detailed, Though this project is very great difficult to me but in a way that is very constructive——I learned a lot from it.
Keywords/Search Tags:Association Rules, Apriori Arithmetic, Multi-value Property, Data Mining, Business Intelligence
PDF Full Text Request
Related items