A Generic Approach for Relative Regular-Frequent Pattern Mining in Retail Supermarket

Regular-Frequent itemsets mining is a challenging data mining concept to discover interesting and useful patterns in retail transactional market databases. It offers the most extreme advantages to a retailer for product placements, segmentation, layout designs, promoting strategies and different vending decisions to enhance customer satisfaction as well as increase sales. Most of the existing studies consider only the frequent patterns that strictly satisfy the user defined maximum regularity threshold level for all of its periods in the transactional database. Many frequent patterns satisfy the threshold level in almost all of its periods but these patterns are not considered regular-frequent pattern due to the strict periodicity measure. The retail market is highly seasonal. Many seasonal itemsets regularly appears at particular time intervals in a transactional database, those aren't regular frequent at all. In this paper, we propose a generic and flexible approach for mining relative seasonal regular-frequent patterns in retail transactional databases. The dataset has been generated from transaction slips in a large Supermarket of Bangladesh. Our results represent the superiority over the traditional systems in terms of discovering many hidden useful patterns in the transactional database for making easy merchandising decisions