The Algorithm Research Of Associative Classification Rule Mining And Its Application In Medical Image Data Mining

Classification is the key tasks of data mining. And the classification based on class association rules as a new method of classification has attracted attention from academic research community and industry field since the first algorithm of Classification Based on Association rules was introduced in 1998.With the development of computer, there is a great deal of database of medical information. Owing to medical technology has feature of experimental and statistic, the data mining has widely and practical prospects. Therefore, data mining for medical image has been a flourishing research area between medical and computer science now. The research on medical image mining has just started. It is significant to research and find powerful data mining algorithms that can handle medical image datasets effectively and efficiently. Further, it is important to diagnosis disease for doctor.This paper aims at achieving more effective associative classifiers based on importance of attribute (IRARC). Compared with other algorithms, this algorithm was proposed to apply the importance of attribute measure to the generation of candidate itemsets. Moreover, in the process of building classifier, a new strategy to rank class association rules in order to discriminate between rules which have identical confidences or supports and to prune rule redundancy and conflicts was proposed. The theoretical analysis and experiment results indicate that this algorithm is of higher application efficiency than CBA algorithm.This thesis studies and analyses the technique and application of associative classification rules mining in medical image mining systematically and deeply. This algorithm was proposed to apply FP-Growth algorithms to solve the short of data mining which involve a great many of long pattern, strong pattern and low threshold. And secondly, applying a new strategy to rank class association rules in process of built a classifier. Experimental results show that the method is efficient and accurate.