Table 3 Average validity of the physicochemical and fingerprint based models
From: Maximizing gain in high-throughput screening using conformal prediction
Confidence level | ||||
|---|---|---|---|---|
90% | 80% | 70% | 60% | |
Physiochemical | ||||
Validity train active | 0.928 | 0.833 | 0.728 | 0.631 |
Validity train inactive | 0.910 | 0.813 | 0.715 | 0.614 |
Validity test active | 0.922 | 0.818 | 0.718 | 0.615 |
Validity test inactive | 0.907 | 0.811 | 0.714 | 0.615 |
Fingerprint | ||||
Validity train active | 0.976 | 0.896 | 0.771 | 0.627 |
Validity train inactive | 0.949 | 0.888 | 0.809 | 0.694 |
Validity test active | 0.972 | 0.895 | 0.766 | 0.610 |
Validity test inactive | 0.943 | 0.884 | 0.810 | 0.714 |