Model and feature selection in microarray classification

Microarray classification has a broad variety of biomedical applications. Support vector machines (SVMs) have emerged as a powerful and popular classifier for microarray data. At the same time, there is increasing interest in the development of methods for identifying important features in microarray data. Many of these methods use SVM classifiers either directly in the search for good features or indirectly as a measure of dissociating classes of microarray samples. The present study describes empirical results in model selection for SVM classification of DNA microarray data. We demonstrate that classifier performance is very sensitive to the SVM´s kernel and model parameters. We also demonstrate that the optimal model parameters depend on the cardinality of feature subsets and can influence the evolution of a genetic search for good feature subsets. The results suggest that application of SVM classifiers to microarray data should include careful consideration of the space of possible SVM parameters. The results also suggest that feature selection search and model selection should be conducted jointly rather than independently.