A New Joinless Apriori Algorithm for Mining Association Rules

In this paper, we introduce a new approach to implementing the apriori algorithm in association rule mining. We show that by omitting the join step in the classical apriori algoritm, and applying the apriori property to each transaction in the transactions database, we get the same results. We use a simulation study to compare the performances of the classical to the joinless algorithm under varying conditions and draw the following conclusions: (1) the joinless algorithm offers better space management; (2) the joinless apriori algorithm is faster for small, but slower for large, average transaction widths. We analyze the two algorithms to determine factors responsible for their relative performances. The new approach is demonstrated with an application to web mining of navigation sequences.