Multinomial event naive Bayesian modelling for SAGE data classification (Q2271701)

Different approaches to classification of serial analyses of gene expression (SAGE) data are compared on real data of human cancerous tissues (brain and breast). SAGE measures expression levels of different genes in a tissue. Naive Bayes classification based on the multinomial event model and \(k\)-nearest neighbours classification with information distance and Wilcoxon rank sum test are discussed. The authors' conclusion is that the Bayesian approach is promising for accurate classification of SAGE data and is more efficient than \(k\)-NN classification.