Skip to main content
Figure 2 | BMC Bioinformatics

Figure 2

From: Challenges in microarray class discovery: a comprehensive examination of normalization, gene selection and clustering

Figure 2

Difference between data sets. Boxplots showing the aRand values for the five data sets. A. The distribution of aRand values for each data set over all 2780 cluster analysis approaches. B. The same distribution, but based only on the 288 cluster analyses that remained after the reduction of cluster analyses. C. The complement to B i.e., distributions based on the 2492 cluster analyses that were removed. The distribution of aRand values for random classifications for each data set is indicated in the plots by horizontal lines representing the median and 95-percentile of the aRand values (dashed and dotted, respectively).

Back to article page