Table 3:
Accuracies of treatment selection in 1000 test cases using the proposed technique for the case of three treatments and four responses, using weights τ1 = 0.4, τ2 = 0.3, τ3 = 0.2, and τ4 = 0.1, for responses 1, 2, 3, and 4, respectively.
Error dist. | Sample size per group | Error dist. parameter | Error correlation | ||
---|---|---|---|---|---|
ρ = 0.1 | ρ = 0.5 | ρ = 0.9 | |||
Normal | N = 50 | σN = 0.1 | 826 | 834 | 839 |
σN = 0.2 | 754 | 782 | 800 | ||
σN = 0.3 | 676 | 679 | 672 | ||
N = 100 | σN = 0.1 | 856 | 877 | 904 | |
σN = 0.2 | 763 | 816 | 848 | ||
σN = 0.3 | 693 | 706 | 738 | ||
DE | N = 50 | σD = 0.1 | 800 | 830 | 874 |
σD = 0.2 | 743 | 759 | 769 | ||
σD = 0.3 | 687 | 667 | 657 | ||
N = 100 | σD = 0.1 | 844 | 875 | 912 | |
σD = 0.2 | 753 | 810 | 839 | ||
σD = 0.3 | 699 | 689 | 728 |