Figure 3
From: Individualized markers optimize class prediction of microarray data
Procedure for building a dendrogram. Using the distance metric shown in equation 1, we determine a similarity score between two samples that reflects the number of common votes provided by the same informative genes. A. Based on this score, samples T1 and T2 in the training set form a bifurcation note representing their respective classes. B. Unseen samples S1 and S2 are placed on this node according to their similarity with T1 and T2.