Discovery of Potential High Utility Itemset from Uncertain Database using Multi Objective Particle Swarm Optimization Algorithm

Abstract

In recent decades, Internet of Things devices have grown in popularity across a wide range of industries and uses. As a result, vast amounts of data are created and generated. Despite the fact that the collected data contains a great quantity of crucial information, most current and general pattern mining algorithms simply analyses a single item and exact information to identify the needed data. Because the amount of data gathered is so huge, it is vital to identify meaningful and updated data in a short period of time. In this paper, we use a multi-objective evolutionary framework to effectively mine the interesting Potential High Utility Itemset (PHUI) in a limited period, with the majority of items being PHUI utility and uncertainty. In an unpredictable context, the benefits of the proposed model (dubbed MOPSO-PHUIM) can identify lucrative PHUIs without pre-defined threshold values (i.e., minimal utility and minimum uncertainty). To illustrate the efficiency of the created MOPSO-PHUIM, two encoding techniques are also taken into account. Using the developed MOPSO-PHUIM model for decision-making, a set of non-dominated PHUIs may be found in a short amount of time. Studies are then carried out to demonstrate the utility and performance of the built MOPSO-PHUIM model in terms of velocity, hyper volume, and the different result discovered when compared to generic techniques.