Table 4:
Comparison of different types of features used to train the models for obesity. The mean and standard deviation are recorded for different evaluation metrics after five runs of 5-fold cross-validation. The best performances are highlighted in bold.
| Microbial Abundances | k-mer Abundances | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Accuracy | Precision | Recall | F1-Score | AUC | Accuracy | Precision | Recall | F1-Score | AUC | |
| SVM | 0.6374 (0.0008) |
0.6374 (0.0008) |
1.0000 (0.0000) |
0.7768 (0.0012) |
0.5133 (0.0361) |
0.6154 (0.0271) |
0.6923 (0.0199) |
0.7229 (0.0296) |
0.7031 (0.0202) |
0.5993 (0.0174) |
| RF |
0.6480 (0.0112) |
0.6512 (0.0034) |
0.9675 (0.0182) |
0.7764 (0.0087) |
0.6416 (0.0062) |
0.6139 (0.0161) |
0.6733 (0.0054) |
0.7786 (0.0272) |
0.7170 (0.0155) |
0.5937 (0.0268) |
| XGBoost | 0.6352 (0.0241) |
0.6749 (0.0112) |
0.8277 (0.0366) |
0.7407 (0.0205) |
0.6055 (0.0241) |
0.6169 (0.0261) |
0.6818 (0.0116) |
0.7614 (0.0427) |
0.7145 (0.0253) |
0.5979 (0.0196) |
| gcForest | 0.6404 (0.0125) |
0.6553 (0.0094) |
0.9247 (0.0163) |
0.7644 (0.0082) |
0.6495 (0.0148) |
0.6365 (0.0242) |
0.7042 (0.0194) |
0.7470 (0.0282) |
0.7211 (0.0184) |
0.6186 (0.0337) |
| AutoNN | 0.6238 (0.0072) |
0.6432 (0.0024) |
0.9299 (0.0247) |
0.7572 (0.0077) |
0.6031 (0.0127) |
0.5972 (0.0124) |
0.6665 (0.0106) |
0.7525 (0.0178) |
0.70 01 (0.0108) |
0.5666 (0.0135) |