Enhancing sample classification for microarray datasets using genetic algorithm

Microarray is a high throughput technology that allows uncovering of thousands of genes concurrently. To conduct any biological test like disease prediction and classification in medical field, among the large amount of genes presented in gene expression data, only some particular amount of genes is effective for performing diagnostic test. A Supervised attribute clustering is used to find such initial co-expressed gene groups of clusters whose joint expression is strongly related with the class labels. The Mutual Information incorporates the information of sample categories to measure the similarity between attributes by sharing the information between each attributes. Thus the redundant and irrelevant attributes are eliminated. After forming the clusters, the GA is used to find the optimal feature so as to increase the class separability. Using this method, the diagnosis can be made easier and effective. The predictive accuracy is estimated using three classifiers such as K-nearest neighbors, naive bayes and Support Vector machine. Thus the overall approach provides excellent predictive capability for accurate medical diagnosis.