Skip to main content

Table 1 Performance comparison of different algorithms for side effect prediction

From: Learning important features from multi-view data to predict drug side effects

 Sample-AUCMacro-AUCMicro-AUCLRAPCoverage errorRanking loss
L1SVM0.8524 ± 0.00100.6196 ± 0.00560.8328 ± 0.00100.1941 ± 0.00132671 ± 110.1483 ± 0.0010
L1LOG0.8612 ± 0.00100.6191 ± 0.00590.8418 ± 0.00100.2018 ± 0.00152562 ± 150.1394 ± 0.0010
PCR0.8824 ± 0.00040.5034 ± 0.00330.8670 ± 0.00060.1890 ± 0.00102666 ± 160.1233 ± 0.0004
SCCA-chem0.8500 ± 0.00190.5731 ± 0.00450.8181 ± 0.00190.4008 ± 0.00322960 ± 230.1507 ± 0.0020
SCCA-domain0.9144 ± 0.00070.6260 ± 0.00550.8922 ± 0.00070.4757 ± 0.00112547 ± 130.0863 ± 0.0008
SCCA-GO0.8911 ± 0.00150.6160 ± 0.00570.8579 ± 0.00150.4509 ± 0.00172789 ± 250.1097 ± 0.0015
SCCA-expression0.9076 ± 0.00070.5159 ± 0.00310.8878 ± 0.00070.4488 ± 0.00052607 ± 90.0941 ± 0.0007
SCCA-target0.9174 ± 0.00050.6159 ± 0.00510.8968 ± 0.00050.4692 ± 0.00072490 ± 110.0834 ± 0.0005
Kernel regression0.9185 ± 0.00050.6134 ± 0.00530.8992 ± 0.00050.4766 ± 0.00072448 ± 80.0821 ± 0.0005
LRSL-chem0.9179 ± 0.00030.5595 ± 0.00340.8976 ± 0.00040.4614 ± 0.00052583 ± 100.0867 ± 0.0005
LRSL-domain0.9285 ± 0.00050.6470 ± 0.00500.9104 ± 0.00070.4821 ± 0.00052174 ± 140.0719 ± 0.0005
LRSL-GO0.9290 ± 0.00070.6441 ± 0.00430.9068 ± 0.00100.4924 ± 0.0002255 ± 150.0714 ± 0.0007
LRSL-expression0.9203 ± 0.00040.5131 ± 0.00130.9008 ± 0.00050.4565 ± 0.00062198 ± 110.0801 ± 0.0004
Multi-LRSL0.9295 ± 0.0000.6568 ± 0.00570.9118 ± 0.00090.4845 ± 0.00062160 ± 130.0709 ±0.0006
  1. The metrics are denoted as \(\hbox {mean}\pm \hbox {standard deviation}\) . The method taking different types of features as input is indicated in the form of ’method name-feature type’
  2. The best metric values are in italics, and the difference between the best value and the second best value of each metric is significant (student t-test, \(\hbox {p-value}<0.05\))