TKAR: Efficient Mining of Top-k Association Rules on Real—Life Datasets

Abstract

Data mining is an important facet for discovering association rules among the biggest scope of itemsets. Association rule mining (ARM) is one of the techniques in data processing with the two sub processes. One is identifying frequent itemsets and the other is association rule mining. Frequent itemset mining has developed as a major issue in data mining and assumes an essential part in various data mining tasks, for example, association analysis, classification, etc. In the structure of frequent itemset mining, the outcomes are itemsets which are frequent in the entire database. Association rule mining is a basic data mining task. Researchers developed many algorithms for finding frequent itemsets and association rules. However, relying upon the choice of the thresholds, present algorithms become very slow and produce a greatly large amount of outcomes or generates few outcomes, omitting usable information. Furthermore, it is well-known that an expansive extent of association rules produced is redundant. This is truly a significant issue because in practice users don’t have much asset for analyzing the outcomes and need to find a certain amount of outcomes within a limited time. To address this issue, we propose a one of a kind algorithm called top-k association rules (TKAR) to mine top positioned data from a data set. The proposed algorithm uses a novel technique for generating association rules. This algorithm is unique and best execution and characteristic of scalability, which will be a beneficial alternative to traditional association rule mining algorithms and where k is the number of rules user want to mine.