The use of Histogram Analysis to Support Fast Selection of Predictive Features for Data Stream Mining

A data stream is a real-time, continuous, ordered sequence of items. Many stream mining methods for fast data selection and modeling from large quantities of stream data have been proposed in the literature. Histograms are commonly used in various data stream mining activities such as exploratory data analysis and pre-processing of training data for classification algorithms. This paper reports on an exploratory study of the use of histograms as a basis for fast selection of relevant features for predictive classification. Histograms are analysed using simple statistical measures in order to establish feature relevance. Experimental results demonstrate that the proposed method has a high potential of selecting relevant features which result in high predictive performance.