Improving Performance in Mining Algorithms

Abstract

In Frequent pattern mining, we have to generate and examine a large number of intermediate subsequences. So it‟s a difficult data mining problem with broad applications. Frequent Pattern Mining has been at the core of the field of data mining for the past many years. Frequent Pattern Mining is the task of finding sets of items that frequently occur together in a dataset. With the higher growth of data, users need more relevant and sophisticated information, which may be lying hidden in the data. Data mining is often described as a discipline to find hidden information in a database. It involves different algorithms and techniques to discover useful knowledge lying hidden in the data. Most of the previously developed sequential pattern mining methods, such as Generalized Sequential Pattern algorithm (GSP), explore a candidate generation-and-test approach [1] to reduce the number of candidates to be examined. However, this algorithm may not be efficient in mining sequence databases which contains a number of numerous patterns and/or long patterns. Since it produces a large number of candidate sequences and does a lot scan of database. Algorithms like Prefix Span[2], finds frequent pattern subsequences using pseudo/projected database and BI-Directional Extension (BIDE)[3] finds frequent closed pattern mining for single data elements , none of the algorithm does closed pattern mining for multiple data sets. In this thesis, we propose an efficient algorithm for finding the closed frequent patterns using for multiple data Item Set thus doing a general database mining, which is not limited to the number of datasets an element has. It removes the redundant pattern and gives less number efficient pattern.