Skip to main content

Table 7 The results of ablation study of using different components of MetaMLAnn by cross validation on New York and Boston datasets at genus level

From: Prediction of microbial communities for urban metagenomics using neural network approach

Evaluation metric
  Precision Recall F1 score Ranking loss
New York dataset
Bshared only 0.7388 0.4986 0.5952 0.1299
Bi only 0.7339 0.5379 0.6204 0.0765
Bi+Bshared 0.7456 0.5325 0.6212 0.0682
Boston dataset
Bshared only 0.6002 0.5920 0.5890 0.1896
Bi only 0.7574 0.5660 0.6428 0.1316
Bi+Bshared 0.7674 0.6706 0.7095 0.1270
  1. Higher precision, recall, F1 score, and lower ranking loss indicate better performance. Bold entries indicate best performance among different methods