Fig. 5

Average performance of feature selection algorithms. Average performance is obtained using 9 combinations of the synthetic microarray model parameters k and mean type with fixed ρ₀ and ρ₁, where performance is defined to be the average number of markers identified as good plus the average number of non-markers identified as bad over 500 iterations: a ρ₀=0.1,ρ₁=0.1, b ρ₀=0.5,ρ₁=0.1, c ρ₀=0.9,ρ₁=0.1, d ρ₀=0.1,ρ₁=0.5, e ρ₀=0.5,ρ₁=0.5, f ρ₀=0.9,ρ₁=0.5, g ρ₀=0.1,ρ₁=0.9, h ρ₀=0.5,ρ₁=0.9, and i ρ₀=0.9,ρ₁=0.9