Figure 3
From: Preprocessing of gene expression data by optimally robust estimators
Tukey's biweight versus rmx estimator. Mean standard deviation (SD) versus mean log expression for Tukey's biweight and the rmx estimator for s ∈ [0,0.05]. The curves were determined by smoothing the resulting scatterplots which include SDs and mean log expressions for each gene not spiked-in. As the variability in terms of mean SD for the rmx estimator is clearly smaller especially for HGU95A, these plots indicate an improvement of the accuracy of MAS 5.0 by using the rmx estimator instead of Tukey's biweight.