Fig. 1
From: Transferring genomics to the clinic: distinguishing Burkitt and diffuse large B cell lymphomas
Performance of different machine learning algorithms with two previous data sets. Overall error rates (tenfold cross-validation within the data set GSE4732_p1, GSE4475_strict and GSE4475_wide, respectively) for the binary classification problem using a range of machine learning methods (LibSVM, SMO, MultilayerPerceptron, Random Forest, Function Tree, LMT, BayesNet, NaiveBayes, J48 and REP Tree, all implemented in Weka machine learning tool) with default parameters. In GSE4475 we consider two possible definitions of BL, strict (cases for which the authors give a BL probability of >0.95) and wide (BL probability >0.5). Classifiers are tested with the gene sets employed in the original papers for these data sets (214 genes for GSE4732_p1, 58 genes for GSE4475 strict and wide definition)