Skip to main content

Table 6 Predictive results of the models built by removing one of the selected features

From: Protein-protein interface hot spots prediction based on a hybrid feature selection strategy

Feature removed Accuracya Specificity Recall Precision F-measure
BsRASA 0.818/0.684 0.815/0.657 0.807/0.750 0.746/0.477 0.775/0.583
CNSV_REL1 0.786/0.632 0.783/0.582 0.790/0.750 0.710/0.429 0.748/0.545
DtASA 1/2 0.818/0.684 0.826/0.642 0.807/0.786 0.758/0.478 0.781/0.595
UpASA 0.786/0.632 0.815/0.627 0.742/0.643 0.730/0.419 0.736/0.507
BtRASA 0.805/0.705 0.837/0.672 0.758/0.786 0.758/0.500 0.758/0.611
B factor 0.825/0.705 0.815/0.658 0.839/0.821 0.754/0.500 0.794/0.622
  1. aFor columns 2–6, the values above the slashes are for cross validation, and the values under the slashes are for the independent test set