Improvement of Apriori Algorithm Based on Vector and Vertical Array

In the data mining method of association analysis, the classic Apriori algorithm of discovering frequent item sets may multiple scanning the source database, produce a large number of candidate and repeatedly pattern matching, which leads to low time efficiency of the algorithm. Based on the analysis of the array based algorithm, an improved algorithm is proposed in this paper. The main idea is to scan the source database once and use vector arrays and vertical arrays to represent the transactions, improve the strategy of the join step and the prune step when candidate frequent(k+1)-item sets were generated from frequent(K)-item sets as well as the pattern matching strategy. The experimental results show that the time complexity of the improved algorithm is reduced greatly.