Table 3 Performance evaluation of 9 models. ACC: accuracy; SEN: sensitivity; SPE: specificity; AUC: area under the ROC curve; \(dis{{t}_{\text {AVG}}}\): the average distance between the counterfactual example and the original example.
Metric | FLE vs. NC | TLE vs. NC | SZ vs. NC | ||||||
|---|---|---|---|---|---|---|---|---|---|
DT | AB | RF | DT | AB | RF | DT | AB | RF | |
ACC(%) | 84.10 | 81.82 | 82.03 | 85.37 | 82.93 | 80.49 | 82.74 | 84.09 | 80.49 |
SEN(%) | 85.71 | 71.43 | 89.47 | 92.31 | 85.71 | 82.35 | 73.68 | 75.18 | 72.22 |
SPE(%) | 81.82 | 88.24 | 73.91 | 70.59 | 70.26 | 73.68 | 87.50 | 88.24 | 81.25 |
AUC(%) | 84.16 | 81.37 | 82.73 | 87.23 | 83.60 | 80.76 | 82.84 | 82.93 | 79.59 |
\(dis{{t}_{\text {AVG}}}\) | 0.798 | 0.857 | 0.832 | 0.261 | 0.287 | 0.430 | 0.170 | 0.169 | 0.418 |
