TABLE 3.
Overall rates of correct classification (ARCC) for different reductions in numbers of drugs and species classifications for three DA methods: results from cross-validation analysis
| Approach (probability due to random chance) | % of isolates correctly classified by DA method, best group for identificationa
|
||
|---|---|---|---|
| Linear | Quadratic | Epanechnikov | |
| Twelve drugs; eight species (12.5%) | 28.7,* swine (49) | 25.6, wildlife (75) | 27.5, humans (67) |
| Reduced drugs; eight species (12.5%) | 30.8, swine (50) | 27.7, wildlife (75) | 37.3,* wildlife (89) |
| Twelve drugs; six species (16.7%) | 35.0,* swine (59) | 33.1, wildlife (80) | 33.8, humans (67) |
| Reduced drugs; six species (16.7%) | 37.1, swine (57) | 35.0, wildlife (78) | 54.5,* humans (100) |
| Twelve drugs; five species (20%) | 33.6, livestockb (67) | 32.7, wildlife (83) | 42.3,* humans (100) |
| Reduced drugs, five species (20%) | 36.3, livestockb (67) | 35.4, wildlife (80) | 51.5,* wildlife (89) |
| Twelve drugs; four species (25%) | 32.6, food animalsc (82) | 36.8, wildlife (84) | 48.7,* humans (100) |
| Reduced drugs; four species (25%) | 39.0, food animalsc (80) | 40.7, wildlife (81) | 67.9,* humans (100) |
a
*, best model by approach; the value in parentheses following the best group for identification is the percentage of isolates correctly identified in the group.
b
The livestock group is a combination of beef and dairy cattle, sheep, and swine.
c
The food animal group is a combination of beef and dairy cattle, sheep, swine, and poultry.