Feature Clustering Towards Gene Selection

Microarray datasets usually contain a small size of samples versus a large number of genes as features. Feature selection is used for selecting a subset of features that lead to better performance of the classifier. However, most feature selection algorithms suffer from high computational complexity when applied to microarray datasets. To tackle this problem, we incorporate feature clustering in the gene selection. Results in this paper show that reducing the size of datasets using our approach, significantly reduces the computational time of feature selection algorithms while maintaining or improving the prediction power of the classifier.