Table 4.
Evaluation of the generated candidate rules. The testing dataset was used to validate the corresponding set of rules. For instance, the fungus testing dataset was used to evaluate the set of rules generated from the fungus training dataset.
| A | B | C | D | E | |
| precision | 71% | 76% | 87% | 88% | 77% |
| confidence | 69% | 74% | 85% | 85% | 75% |
| coverage* | 43% | 38% | 60% | 54% | 56% |
A: actinobacteria B: bacillales C: fungi D: nematode + arthropoda E: viridiplantae
*: coverage = the hit ratio of testing data