Positive And Negative Association Rules Mining Research And Applications

With the rapid development of information society, information has been sharped growth in the number, diverse sources, and higher complexity. The application requirement about information becomes more and more geater. The mined data has been analyized and understood, which has been shown that data mining is one of the most active area of current researches, this methold can reveal hidden useful informations. The interassociation of data has been revealed by asociation rules mining, which has important theoretical values and broad the application prospects.With items of data, the association rules have positive and negative rule. At new, positive association rules have been widely concerned, but association rules with negative items are not given sufficient attentions. However, in many practical application areas, the negative factors are also very important sources of information, so it is necessary to study the negative association rules.This thesis brings forward a new algorithm, which is based on the study of the traditional algorithm of positive association rules and the rising algorithm of negative association rules. The new algorithm is mining frequent itemsets with negative items, based on the frequent pattern tree. Similar to frequent pattern tree of FP-growth algorithm, the algorithm compress negative items into FP-Tree which is same to the positive items, so the advantages are no need to scan the database repeatedly and not generate large amounts of candidate sets. Using sequential structure store FP-Tree, the algorithm has highly time efficiency and space efficiency. Without produce conditional pattern subtree, the algorithm reduces all frequent itemsets for one inccording to row operation of the path-matrix and greatly shorts the execution time. At then same time, this thesis introduces intrestingess as the third parameter of associaton rules, which has been used to prune no intrest rules. Experiments show that the algorithm has better efficiency than the existing similar mining algorithms.Finally, the new aglorithm has been applied to Adult dataset of questionnaire survey. According to analysis the mining association rules, the importance of negative associtation rules has been proved.