Skip to main content

Table 4 Statistical significance of five methods over eighteen test-cases determined through Friedman tests with Hommel’s post-hoc analysis at α=0.05

From: Decoy selection for protein structure prediction via extreme gradient boosting and ranking

Top Basins Method Average Rank p value p Hommel
B1 PR 3.889 2.101E-6 0.0125
  Basin-Size 3.306 2.76E-4 0.0167
  PR+PC 3.306 2.76E-4 0.025
  Basin-Size+Energy 3.11 0.001 0.05
  ML-select 1.389 - -
B1−2 PR 4.028 5.53E-7 0.0125
  Basin-Size 3.417 1.19E-4 0.0167
  PR+PC 3.139 8.99E-4 0.025
  Basin-Size+Energy 3.028 0.002 0.05
  ML-Select 1.389 - -
B1−3 PR 3.833 7.47E-6 0.0125
  Basin-Size 3.444 1.83E-4 0.0167
  PR+PC 3.306 5.04E-4 0.025
  Basin-Size+Energy 2.944 0.005 0.05
  ML-Select 1.472 - -
  1. The best method is marked with an asterisk (*), while the boldface presents the significance of the respective method when compared with the best method