Mining for Enthalpy-Based Average High-Utility Patterns with Tighter Upper Bounds

Mining for high average-utility item sets in a quantitative database is developing the traditional problem of frequent item data mining, with several practical applications. However, such utility measurements for patterns in mining have a drawback when dealing with extended patterns. This article discusses mining using an enthalpy-based high average-utility item set. First, the input database was preprocessed to ensure that the mining process ran smoothly. The total utility is first estimated during pre-processing, and then the threshold is used to determine the minimum average utility. We then use the looser upper bound (lub) and revised tighter upper bound (rtub) models to arrange the moderate utility upper bounds in ascending order. The lub model considers the residual highest utility in transactions to lower the upper bound on the utility of item sets. The rtub model ignores insignificant items in transactions to further reduce the upper bound. Then, a conditional tree is constructed based on the knowledge of the sorted upper bounds. Finally, an enthalpy-based pruning algorithm excludes unsuitable applications early on, resulting in improved mining in high average-utility patterns. The preliminary results indicate that our proposed methodology surpasses current runtime, memory consumption, and scalability strategies.