Study On The Interest-based Association Rules Algorithm

Data mining is a process which extracts the unknown, valuable pattern or rule from large amounts of data. Association rules mining is an important branch of data mining .Mining association rules in databases has received much attention recently. The technology is widely used in the commercial, financial, scientific research and so on. Apriori algorithm is often used in association rule mining, the two thresholds: support and confidence are both used to evaluate the rule is a strong rule or not in association rules mining. The strong association rules for decision-making are helpful. The study shows that the support is widespread in statistics yet the confidence can't be very good to evaluate the rule's credibility. The conditional probability covers up some of the associated properties. In this paper we introduce the method of measure data association : correlation coefficientρto improve the traditional algorithm.We will use interest as the first threshold, which can effectively reflect the characteristics of data correlation : direction and strength of correlation associated and avoid the embarrassment brought by the support as the first threshold. If the support is too low, the redundancy will increase, if not, a large number of effective rules will be filtered. The rules from improved algorithm are more accurate and the meaning is clearer. In addition, we the Interest-based Association Rules Algorithm improve the efficiency of the mining. We will try to improve the mining efficiency by reducing the cost of data storage and data access in the process.