Skip to main content

Table 4 Summary of the top five features which are most important for the classification of EFR according to the GMLVQ and RF method

From: Application of an interpretable classification model on Early Folding Residues during protein folding

  GMLVQ Random Forest
Feature Rank Influence score Rank Influence score
PlipHpCL 1 0.159 3 1.370
LF 2 0.127 2 1.403
PlipBN 3 0.063 15 0.900
SecSize 4 0.059 19 0.854
e 5 0.042 4 1.332
PlipCL 7 0.012 1 1.700
ConvCC 23 0.009 5 1.223
  1. Importance scores for the RF were computed by the MATLAB implementation. Influence scores are in arbitrary units, higher values occur for features important for class discrimination. The values of GMLVQ and the predictor importance values are method-specific and not directly comparable; therefore, the ranks of the top five features are given