Skip to main content

Table 3 The improvement ratio (IR) values of different classifiers on different datasets

From: Selecting high-quality negative samples for effectively predicting protein-RNA interactions

Dataset SVM RF NB
  SE SP GM ACC SE SP GM ACC SE SP GM ACC
SO 1:1 20.2 –12.8 2.4 3.1 4.5 5.8 5.1 5.1 107.2 0 43.9 23.1
SO 1:2 2.5 0.23 1.4 1.0 4.4 1.0 2.7 2.1 48.4 3.0 23.6 11.4
SO 2:1 0.6 3.7 2.1 1.6 0.6 6.0 3.2 2.3 12.1 4.2 8.1 8.5
HOMO 1:1 86.5 –35.4 9.8 5.3 42.1 17.2 29.0 30.8 –52.2 120.4 2.6 5.0
HOMO 1:2 109.2 12.3 53.3 25.2 65.4 29.0 46.1 35.5 –59.9 302.2 26.9 54.3
HOMO 2:1 2.3 94.5 41.1 15.7 19.1 63.1 39.4 26.8 –30.3 40.3 –1.1 -4.6
MUS 1:1 249.5 –63.4 13 68.7 372.2 –18.7 95.9 53.9 –21.1 299.5 77.5 21.8
MUS 1:2 760.4 7.8 204.5 29.3 751.4 –11.5 174.5 16.7 32.6 –28.1 –2.4 6.7
MUS 2:1 2.7 497.8 147.8 11.7 3.8 286.9 100.4 31.9 –18.0 299.1 80.9 0