Figure 1
From: Speeding up the Consensus Clustering methodology for microarray data analysis
An example of number of cluster prediction with the use of Consensus and FC. The experiment is derived with the Leukemia dataset as input, with the use of the K-means-A clustering algorithm. (i) The plot of the CDF curves as a function of k, obtained by Consensus with H = 250 and p = 80%. For clarity, only the curves for k in [2, 13] are shown. It is evident that there are increasing values of the area under the CDF for increasing values of k. The flattening effect in the growth rate of the area is evident for k ≥ k* = 3. (ii) The plot of the corresponding Δ curve for k in [2, 30], where the flattening effect indicating k* is evident for k ≥ k* = 3. (iii) The plot of the CDF curves, obtained by FC with H = 250 and p = 80%, in analogy with (i). (iv) The plot of the Δ curve, obtained by FC with H = 250 and p = 80%, in analogy with (ii).