Research on Top-k Association Rules Mining Algorithm Based on Clustering

Abstract

Until now, the association rule mining algorithm is still one of the core issues in the field of big data research. At present, there are many research related to association rule mining algorithms, mainly focusing on how to find frequent item sets and how to tailor rules, and based on this direction, there have been many classic algorithms, such as Apriori and FP-growth algorithms. However, most of the above algorithms process and analyze the entire data set, so that although related algorithms can be used to obtain the results, the results are not meticulous. To solve this problem, this paper proposes an algorithm, which first uses the k-means algorithm to perform cluster analysis on the data set to generate a specific number of clusters. Then, in different clusters, the association rules in each cluster are found by the improved Top-k algorithm combined with the correlation coefficient. By integrating clustering and the improved Top-k algorithm, the data set can be analyzed directionally to improve the accuracy and efficiency of the whole algorithm. And the final experiment shows that compared with the original algorithm, the running time is shortened by 14%.