A marketing solution for cross-selling by high utility itemset mining with dynamic transactional databases

Utility-base mining has emerged as an alternative approach for frequent itemset mining because it reflects statistical correlation and semantic significance among the items. Dynamic databases require a provision for adding new data to the previous analysis results to handle existing and newly generated data efficiently. This paper introduces a system which uses HUPID-Tree structure to maintain the information about the database and patterns and it is updated only with the incremented data. It reduces the time overhead of rescanning the database from the beginning. High utility itemsets (HUIs) i.e. the desirable patterns mined from the HUPID-Tree will be used for generating rules. Cross selling profit of each rule will be estimated with the help of an objective function i.e. the rule utility function. Cross selling is the practice of selling among the established customers. It uses items in the consequent part of a rule for recommendation and provides future profit information with the application of a rule. Managers can use this cross-selling profit information to maximize the profit and the itemsets which will be sold in the future will also be the high utility itemsets.