Research Article
A Microcosmic Syndrome Differentiation Model for Metabolic Syndrome with Multilabel Learning
Table 2
Evaluation of prediction results from ML-kNN, kNN, DT, and SVM using physicochemical indexes (PI).
| Evaluation criteria | ML-kNN | kNN | DT | SVM |
| Average precision | 0.714 ± 0.024 | 0.497 ± 0.028 | 0.488 ± 0.036 | 0.554 ± 0.039 | Hamming loss | 0.233 ± 0.021 | 0.297 ± 0.030 | 0.308 ± 0.020 | 0.236 ± 0.028 | Ranking loss | 0.169 ± 0.012 | 0.698 ± 0.053 | 0.678 ± 0.044 | 0.706 ± 0.046 | Coverage | 5.123 ± 0.476 | 7.512 ± 0.894 | 7.866 ± 0.796 | 7.648 ± 0.743 |
|
|
Representing the index in this model is the best compared with others. |