The Applied Research Of Incremental Updating Algorithms For Mining Classification Association Rules

Following the rapid development of database and computer hardware, it is possible to collect plenty of data. To a certain extent, the large amount of data let people easy to describe the characteristics and make decision. But at the same time, it guide people to the awkward position, which was named"data is rich, but knowledge is poor". In order to solve this problem, the data mining emerged in 1980s and became an important technology of information technology.In the multiple branch of data mining, both mining association rules and classification applied research are the most active areas. Because they have the significant similarities in mining strong related project, the technology of mining association rules was used in the application of classification. And found in the study that because of association rules mining can regard the high degree relation of several attributes, it tend to get higher accuracy rate.Based on the brief analysis of the data mining research at home and abroad, this paper introduced the generation, concepts, processes, function and so on. Among them, it focused on an overview of mining association rules, including its definition, the classic algorithms, and incremental updating of association rules.Subsequently, in Chapter Three it demonstrated the need of incremental updating classification association rules. And based on it, the paper raised a new improved algorithm to solve the problems of updating classification rules when the minimum support threshold (minsup) changed. Then further raised two new algorithms to obtain the updating classification rules when the database changed, including adding some new data and deleting the old data.And in Chapter Four, the paper use the data set, which named German Credit Data, from UCI as an instance for the incremental updating classification association rules system design and implementation, further examine these algorithms in Chapter Three.The paper is for application-oriented classification and prediction data mining technology, such as customer relationship management, product sales analysis, to promote such industries as business and financial intelligence, etc. It has a certain degree of guidance and reference significance.