Home // DBKDA 2011, The Third International Conference on Advances in Databases, Knowledge, and Data Applications // View article

An Approach for Distributed Streams Mining Using Combination of Naïve Bayes and Decision Trees

Authors:
Meng Zhang
Guojun Mao

Keywords: data stream mining; distributed streams

Abstract:
Nowadays we have various kinds of data generated at high speed in distributed environment. In many cases, it is difficult or unallowed to gather all the distributed data into a central place for processing. So we have to perform part of the work at the location where data is generated. In this paper, we present an approach for mining distributed data streams by using combination of naïve Bayes and decision tree classifiers. The method takes advantage of both distributed and stream mining characteristics. Each local site uses ensemble classifiers of decision tree to learn a concept of incoming stream and transmits local pattern to a central site. The central site combines the collected patterns to build a global pattern using an attribute-weighted strategy in order to relax the attribute independent constraint of naïve Bayes model. The experiment shows that by using this approach we can get comparable understanding of the global data while reducing transmission load and computation complexity.

Pages: 29 to 33

Copyright: Copyright (c) IARIA, 2011

Publication date: January 23, 2011

Published in: conference

ISSN: 2308-4332

ISBN: 978-1-61208-115-1

Location: St. Maarten, The Netherlands Antilles

Dates: from January 23, 2011 to January 28, 2011