Skip to main content

Table 2 Performance of S-ISOMAP compared with other feature selection methods on the test set

From: Prediction of hot spots in protein–DNA binding interfaces based on supervised isometric feature mapping and extreme gradient boosting

Method SEN SPE PRE F1 ACC MCC AUC
SVM-RFE (19) 0.423 0.763 0.555 0.478 0.625 0.197 0.635
mRMR (30) 0.538 0.711 0.569 0.549 0.642 0.251 0.696
RF-SFS (17) 0.654 0.737 0.629 0.642 0.703 0.388 0.709
VSURF (10) 0.678 0.776 0.672 0.669 0.736 0.431 0.704
S-ISOMAP (3) 0.707 0.819 0.721 0.713 0.768 0.508 0.773
  1. The highest value in each column is shown in bold. The numbers in parentheses represent the feature dimensions after dimensionality reduction