A quasi-linear approach for microarray missing value imputation

Authors:
Yu Cheng;Lan Wang;Jinglu Hu
Affiliations:
Graduate School of Information, Production and Systems, Waseda University, Kitakyushu-shi, Fukuoka, Japan;Graduate School of Information, Production and Systems, Waseda University, Kitakyushu-shi, Fukuoka, Japan;Graduate School of Information, Production and Systems, Waseda University, Kitakyushu-shi, Fukuoka, Japan
Venue:
ICONIP'11 Proceedings of the 18th international conference on Neural Information Processing - Volume Part I
Year:
2011

Citing 3
Cited 0

The nature of statistical learning theory

The nature of statistical learning theory
Missing value estimation for DNA microarray gene expression data: local least squares imputation

Bioinformatics
LIBSVM: A library for support vector machines

ACM Transactions on Intelligent Systems and Technology (TIST)

Quantified Score

Hi-index	0.00

Visualization

Abstract

Missing value imputation for microarray data is important for gene expression analysis algorithms, such as clustering, classification and network design. A number of algorithms have been proposed to solve this problem, but most of them are only limited in linear analysis methods, such as including the estimation in the linear combination of other no-missing-value genes. It may result from the fact that microarray data often comprises of huge size of genes with only a small number of observations, and nonlinear regression techniques are prone to overfitting. In this paper, a quasi-linear SVR model is proposed to improve the linear approaches, and it can be explained in a piecewise linear interpolation way. Two real datasets are tested and experimental results show that the quasi-linear approach for missing value imputation outperforms both the linear and nonlinear approaches.