MFCC: An efficient algorithm for frequent closed cube mining

In this paper, a novel algorithm MFCC for frequent closed cube mining is proposed. There are two phases in MFCC algorithm: First, cut the three dimension dataset into several shoes of two dimension dataset, and use proper frequent closed pattern mining algorithm to mine the slices; second, intersect the results of the slices based on efficient pruning rules, and then find all frequent closed cubes. The advantages of MFCC are: 1) MFCC can use the most efficient frequent olosed pattern mining algorithm according to the characteristic of the dataset; 2) Efficient pruning rules can prune away as early as possible all the branches which can not lead to any frequent closed cubes. As a result, it avoids checking of the closeness of the results, which is necessary in previous work. Extensive experiments are conducted on both real and synthetic datasets. The results show that MFCC outperforms other algorithms.