SPMLS : An Efficient Sequential Pattern Mining Algorithm with candidate Generation and Frequency Testing

Sequential pattern mining is a fundamental and essential field of data mining because of its extensive scope of applications spanning from the forecasting the user shopping patterns, and scientific discoveries. The objective is to discover frequently appeared sequential patterns in given set of sequences. Now-a-days, many studies have contributed to the efficiency of sequential pattern mining algorithms. Most existing algorithms have verified to be effective, however, when mining long frequent sequences in database, these algorithms do not work well. In this paper, we propose an efficient pattern mining algorithm, SPMLS, Sequential Pattern Mining on Long Sequences for mining long sequential patterns in a given database. SPMLS takes up an iterative process of candidate-generation which is followed by frequency-testing in two phases, event-wise and sequence-wise. Event-wise phase presents a new candidate pruning approach which improves the efficiency of the mining process. Sequence-wise phase integrates considerations of intra-event and inter-event constraints. Simulations are carried out on both synthetic and real datasets to evaluate the performance of SPMLS.