Figure 5From: Entropy-based gene ranking without selection bias for the predictive classification of microarray dataSubfigure 5(a): Global model; Subfigure 5(b): Run 40: Estimated 3-fold error curves for E-RFE models trained on feature sets of increasing size and without any use of test data in feature ranking (experimental run 40 of 50 on colon cancer). In each plot, the solid curve is an exponential fit, while the saturating n* is indicated by a dotted line.Back to article page