CPTSW: An Efficient Approach for Mining Static Weighted Frequent Patterns From Data Stream

Abstract

In the recent year data mining is one of the most demanding sectors of computer science which basically deals with discovering frequent patterns by using methodologies, techniques and intelligence tools from databases. As the modern technology is growing rapidly, high volume of data with several features are generated by modern applications. When data set’s flowing velocity is high but applications demand real time analyzing of data depends on immediate features then situation has become more challenging. Several researches has been made in order to assuage the challenges regarding data streams.To find the actual patterns as though the nature of data sets is streams, frequent patterns of those data may be huge and requires further mining. Today’s generation are very much interest in patterns that are significant for them and not just all frequent patterns. Already many researchers work with this topic but they are not enough sufficient. In this thesis we proposing a novel tree based approach, CPTSW-growth which is able to capture the uncertain data streams depending on the importance of applications and only produces significant patterns.