ETP-Mine: An Efficient Method for Mining Transitional Patterns

TL;DR

This paper introduces ETP-Mine, an efficient method that improves the discovery of transitional patterns in transaction databases by pruning candidate items, reducing computational effort and enhancing performance.

Contribution

The paper proposes a modified algorithm that prunes candidate items during frequent pattern generation, significantly reducing the number of patterns needed for transitional pattern discovery.

Findings

Reduces the number of frequent patterns generated.

Improves efficiency of transitional pattern mining.

Validated through extensive simulation tests.

Abstract

A Transaction database contains a set of transactions along with items and their associated timestamps. Transitional patterns are the patterns which specify the dynamic behavior of frequent patterns in a transaction database. To discover transitional patterns and their significant milestones, first we have to extract all frequent patterns and their supports using any frequent pattern generation algorithm. These frequent patterns are used in the generation of transitional patterns. The existing algorithm (TP-Mine) generates frequent patterns, some of which cannot be used in generation of transitional patterns. In this paper, we propose a modification to the existing algorithm, which prunes the candidate items to be used in the generation of frequent patterns. This method drastically reduces the number of frequent patterns which are used in discovering transitional patterns. Extensive simulation test is done to evaluate the proposed method.