Research On Data Stream Of Closed Pattern Mining Based On Variable Slide Window

The technology of frequent closed item sets of data stream mining has been used more extensively, and this makes the researchers take more notes on the technology of frequent closed item sets of data stream mining, especially in business decision-making, knowledge base, it plays a significant role. However, because of the high-speed data streams, massive, diverse, unlimited features, the data stream mining suffers a great challenge. Some researchers use data structures to store data of all patterns, and sliding window mechanism to mining, so the closed frequent pattern mining algorithm based on sliding window is proposed, great progress comes in the field of data stream mining. These early algorithms of frequent closed item sets of data stream mining in the sliding window mechanism consider the more rapid and accurate result, ever less into account time-varying characteristics of data streams, so the algorithms are almost in the environment of the data equal flowing. But in real life, data is often not constant, so that the gap between the results and the practical application of research to the data stream mining has brought new problems. In order to find a solution to these new problems in the data stream mining, this paper proposes an algorithm for mining frequent closed patterns in the data stream based on a variable sliding window, mainly for non-constant data stream mining. For the practical value of data stream mining algorithms, it is very important.In order to mining the data stream better, this article, at first, for the concept of data stream, data stream processing system, data stream and mining frequent pattern, mining frequent closed patterns, understand their theories, their nature and conducted in-depth understanding. Second, the various data stream algorithm for mining frequent closed item sets based on the study, analyzed DSCFI algorithm. Finally, it improves the sliding window mechanism, proposing a variable sliding window and the algorithm DS-stream, with synthetic data sets to experiment and analyze the experimental results. The results show that, DS-stream algorithm for mining frequent pattern in data streams have a good time and space efficiency.