Skip to main content
. Author manuscript; available in PMC: 2011 Apr 19.
Published in final edited form as: J Am Stat Assoc. 2010 Dec;105(492):1358–1375. doi: 10.1198/jasa.2010.ap09250

Figure 5.

Figure 5

Simulation study: ROC curves for the three methods cghMCR (in blue), H-HMM (in red) and BDSAcgh (in black) broken down by prevalence (ω). The vertical axis in each plot is the sensitivity (true positive rate) and the horizontal axis is 1-specificity (false positive rate). The columns correspond to the varying noise levels in the data with leftmost column for τ = 0.1, the middle column for τ = 0.2 and the rightmost column for τ = 0.3. The 5 rows from top to bottom correspond to increasing ω = (0.1, 0.2, 0.3, 0.4, 0.5).