Table 2.
Validation results. k-EMD was tested on a dataset, that includes 34 out- breaks; MinDist, ReD and VOICE were validated earlier on a smaller dataset, that didn’t include one of the outbreaks. For convenience, results for k-EMD contain 2 values - one for the smaller dataset, and one for the entire (34 outbreaks) set of hosts (values in parentheses)
Method | k-EMD | MinDist | MinDistB | ReD | VOICE-D | VOICE-S |
---|---|---|---|---|---|---|
Relatedness sensitivity, % | 80.4 (90) | 90 | 92.9 | 55.3 | 85.2 | 86.8 |
Clustering sensitivity, % | 100 (100) | 100 | 100 | 96.3 | 98.2 | 98.2 |
Direction accuracy, % | 88.7 (90.4) | N/A | N/A | 87.1 | 83.9 | 87.1 |
Source accuracy, % | 80 (81.8) | 50 | 40 | 90 | 80 | 90 |