Table 4. Ranking of each marker according to its importance for predicting Lung Cancer.
Marker | Boosted Regression | Random Forests | ||||
---|---|---|---|---|---|---|
Rank | Median Importance a | Range | Rank | Median Importance a | Range | |
CAL | 1 | 69 | 56–82 | 1 | 10 | 9–10 |
sCD26 | 2 | 16 | 6–31 | 2 | 7 | 6–8 |
EGF | 3 | 13 | 5–24 | 3 | 6 | 6–7 |
sEGFR | 4 | 2 | 0–7 | 4 | 5 | 4–5 |
VEGF | 5 | 0 | 0–4 | 6 | 4 | 4–4 |
HB-EGF | 6 | 0 | 0–1 | 5 | 5 | 4–5 |
a Importance measures report method-specific estimates of the individual contribution to prediction of each evaluated biomarker. Medians values and range over 1000 runs are provided.