Title: Robust classification ensemble method for microarray data

Authors: Dongjun Chung; Hyunjoong Kim

Addresses: Department of Statistics, University of Wisconsin-Madison, WI 53706, USA. ' Department of Applied Statistics, Yonsei University, Seoul 120-749, Korea

Abstract: The objective of this study is to develop an accurate and robust classification ensemble method suitable for microarray data with noises. We proposed an algorithm, pattern match (PM)-bagging, which performs well in accuracy and is robust to noise variables and noise observations. From the experiments with real data set, the performance of the proposed method is found quite comparable and not much degraded even when the data set has noise variables or noise observations, while some other ensemble methods showed degradations of performance. A bias and variance decomposition showed that the success of the proposed method is due to an effective reduction of both bias and variance.

Keywords: classification ensemble; microarray data; robustness; data mining; bioinformatics; decision trees; noise variables; noise observations.

DOI: 10.1504/IJDMB.2011.043032

International Journal of Data Mining and Bioinformatics, 2011 Vol.5 No.5, pp.504 - 518

Received: 19 Feb 2009
Accepted: 24 Dec 2009
Published online: 24 Jan 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article