SEQUENTIAL PATTERN MINING ON MULTIDIMENSIONAL MEDICAL STORE DATA CONSIDERING GENERAL CONSTRAINTS

Abstract

Lots of research and enhancement are being done in area of Sequential Pattern Mining. Previous sequential pattern mining algorithms either mines multidimensional sequences, multidimensional sequences with time-interval or sequences with constraints. In this paper, we propose an algorithm for mining multidimensional sequences considering constraints. These sequences contain time-interval between the occurrences of events (sales of item). Application of this algorithm generates frequent sequences from multidimensional data set. Due to constraints, useful, relevant and less numbers of sequences are generated. The time-interval specified between does not only state the order of occurrence of events but also time-interval between occurrences. Thus generation of such kind of useful and less number of sequences makes analysis and decision making easier. In this paper, we have applied the algorithm on data set from medical store. The experiment shows that the number of sequences generated after application of proposed algorithm is tremendously reduced as compared to those numbers of sequences which are generated using traditional algorithm.