Table 7 The most important features for every algorithm and dataset. For all the cases the three features with the highest permutation importance score are listed, showing the exact scores as well
From: Towards a better understanding of the characteristics of fractal networks
First feature | Second feature | Third feature | ||
|---|---|---|---|---|
Model | DT | diam_logsize (0.285) | hub_connectivity (0.198) | assortativity (0.109) |
RF | hub_connectivity (0.136) | diam_logsize (0.071) | avg_clust (0.020) | |
XGB | diam_logsize (0.250) | hub_connectivity (0.159) | assortativity (0.071) | |
Real | DT | avg_deg (0.107) | hub_connectivity (0.104) | diam_logsize (0.067) |
RF | avg_clust (0.040) | assortativity (0.038) | max_deg_n (0.029) | |
XGB | assortativity (0.112) | hub_connectivity (0.044) | diam_logsize (0.022) | |
All | DT | diam_logsize (0.290) | assortativity (0.118) | hub_connectivity (0.098) |
RF | hub_connectivity (0.086) | diam_logsize (0.082) | avg_clust (0.058) | |
XGB | diam_logsize (0.272) | avg_deg (0.089) | hub_connectivity (0.088) |