Approaches to parallelise Eclat algorithm and analysing its performance for <i>K</i> length prefix-based equivalence classes

Abstract

Frequent item set mining (FIM), is one of the prevalent, well-known methods of data mining and is a topic of interest for researchers in the field of decision making. With the establishment of the period of big data where the data is continuously generated from multidimensional sources with enormous volume, and variety in an almost unrevealed way, transforming this data into valuable knowledge discovery which can add value to the organisations to make an efficient decision making poses a challenge in the present research. This leads to the problem of discovery of the maximum frequent patterns in vast datasets and to create a more generalised and interpretable representation of veracity. Targeting the problems stated above, this paper suggests a parallelisation method suitable for any type of parallel environment. The implemented algorithm can be run on a single computer with multi-core processor as well as on a cluster of such machines.