Utilizing Index‑Based Periodic High Utility Mining to Study Frequent Itemsets

Setiawan, Roy and Le, Dac Nhuong and Rajan, Regin and Subramani, Thirukumaran and Sharma, Dilip Kumar and Ponnam, Vidya Sagar and Kumar, Kailash and Batcha, Syed Musthafa Akbar and Dadheech, Pankaj and Sengan, Sudhakar (2021) Utilizing Index‑Based Periodic High Utility Mining to Study Frequent Itemsets. [UNSPECIFIED]

	PDF Download (3210Kb)
	PDF Download (2495Kb)

Abstract

The potential employability in diferent applications has garnered more signifcance for Periodic High-Utility Itemset Mining (PHUIM). It is to be noted that the conventional utility mining algorithms focus on an itemset’s utility value rather than that of its periodicity in the transaction. A MEAN periodicity measure is added to the minimum (MIN) and maximum (MAX) periodicity to incorporate the periodicity feature into PHUIM in this proposed work. The MEAN-periodicity measure brings a new dimension to the periodicity factor and is arrived at by dividing itemset’s period value by the total number of transactions in that dataset. Further, an algorithm to mine Index-Based Periodic High Utility Itemset Mining (IBPHUIM) from the database using an indexing approach is also proposed in this paper. The proposed IBPHUIM algorithm employs a projectionbased technique and indexing procedure to increase memory and execution speed efciency. The proposed model avoids redundant database scans by generating sub-databases using an indexing data structure. The proposed IBPHUIM model has experimented with test datasets, and the results drawn show that the proposed IBPHUIM model performs considerably better.

Actions (login required)