Mining Regular Patterns in Data Streams Using Vertical Format

The increasing prominence of data streams has been lead to the study of online mining in orderto capture interesting trends, patterns and exceptions. Recently, temporal regularity in occurrencebehavior of a pattern was treated as an emerging area in several online applications like networktraffic, sensor networks, e-business and stock market analysis etc. A pattern is said to be regularin a data stream, if its occurrence behavior is not more than the user given regularity threshold.Although there has been some efforts done in finding regular patterns over stream data, no suchmethod has been developed yet by using vertical data format. Therefore, in this paper wedevelop a new method called VDSRP-method to generate the complete set of regular patternsover a data stream at a user given regularity threshold. Our experimental results show that highlyefficiency in terms of execution and memory consumption.