Skip to main content

Advertisement

Table 5 Average percent loss in gain where training data did not correctly predict maximum gain for the test set

From: Maximizing gain in high-throughput screening using conformal prediction

Cost Total number of partially screened datasetsa Fingerprint based models Physiochemical based models
Number of datasetb %loss Number of datasetb %loss
6 9 6 5.7c 4 2.1
10 10 3 1 3 1.8
14 10 3 1.6 2 0.4
  1. aDatasets where the validation did not indicate that the entire set should be screened for maximum gain
  2. bDatasets where the optimum training set validation setting did not correspond to the maximum test set gain
  3. cFails for dataset 2326: 23.9%. Excluding this result: 2.1%