Analyzing frequent patterns in data streams using a dynamic compact stream pattern algorithm

As a result of modern technology and the advancement in communication, a large amount of data streams are continually generated from various online applications, devices and sources. Mining frequent patterns from these streams of data is now an important research topic in the field of data mining and knowledge discovery. The traditional approach of mining data may not be appropriate for a large volume of data stream environment where the data volume is quite large and unbounded. They have the limitation of extracting recent change of knowledge in an adaptive mode from the data stream. Many algorithms and models have been developed to address the challenging task of mining data from an infinite influx of data generated from various points over the internet. The objective of this thesis is to introduce the concept of Dynamic Compact Pattern Stream tree (DCPS-tree) algorithm for mining recent data from the continuous data stream. Our DCPS-tree will dynamically achieves frequency descending prefix tree structure with only a single-pass over the data by applying tree restructuring techniques such as Branch sort method (BSM). This will cause any low frequency pattern to be maintained at the leaf nodes level and any high frequency components at a higher level. As a result of this, there will be a considerable mining time reduction on the dataset

Level dc:type.qualificationlevel

Doctoral (Level 8)

Year dc:date.issued

2019

Author dc:creator

• Oyewale, A

Language dc:language

en

Identifier

oai:salford-repository.worktribe.com:1329696

OAI identifier oai:identifier

oai:salford-repository.worktribe.com:1329696