IN DATA STREAMS USING CLASSIFICATION AND CLUSTERING DIFFERENT TECHNIQUES TO FIND NOVEL CLASS

Darshana Parikh, Priyanka Tirkha

Abstract: Data stream mining is a process of extracting knowledge from continuous data. Data Stream classification is major challenges than classifying static data because of several unique properties of data streams. Data stream is ordered sequence of instances that arrive at a rate does not store permanently in memory. The problem making more challenging when concept drift occurs when data changes over time Major problems of data stream mining is : infinite length, concept drift, concept evolution. Novel class detection in data stream classification is a interesting research topic for concept drift problem here we compare different techniques for same.

Keywords: Ensemble Method, Decision Tree, Novel Class, Option Tree, Recurring class

DOI: https://doi.org/10.15623/ijret.2013.0208027