Table 1 Model performance comparison between the proposed methods for continuous outcomes with 6 moderate effects
From: Bayesian compositional generalized linear mixed models for disease prediction using microbiome data
m | Proportion\(\ddagger\) | Model | Deviance | \(R^2\) | MSE | MAE |
|---|---|---|---|---|---|---|
100 | 0.2 | BGLM | 1907.1 | 0.670 | 2.930 | 1.363 |
| Â | Â | BCGLM | 1902.9 | 0.671 | 2.919 | 1.360 |
| Â | Â | BCGLMM | 1911.2 | 0.669 | 2.940 | 1.365 |
| Â | 0.5 | BGLM | 1976.7 | 0.714 | 3.104 | 1.402 |
| Â | Â | BCGLM | 1971.7 | 0.715 | 3.091 | 1.399 |
| Â | Â | BCGLMM | 1972.7 | 0.715 | 3.094 | 1.401 |
| Â | 0.7 | BGLM | 2016.5 | 0.740 | 3.203 | 1.424 |
| Â | Â | BCGLM | 2012.1 | 0.741 | 3.192 | 1.421 |
| Â | Â | BCGLMM | 2009.9 | 0.742 | 3.186 | 1.422 |
300 | 0.2 | BGLM | 2147.6 | 0.701 | 3.531 | 1.503 |
| Â | Â | BCGLM | 2130.7 | 0.704 | 3.488 | 1.494 |
| Â | Â | BCGLMM | 2089.8 | 0.712 | 3.386 | 1.478 |
| Â | 0.5 | BGLM | 2450.3 | 0.751 | 4.287 | 1.660 |
| Â | Â | BCGLM | 2430.6 | 0.754 | 4.238 | 1.651 |
| Â | Â | BCGLMM | 2286.1 | 0.775 | 3.877 | 1.584 |
| Â | 0.7 | BGLM | 2621.4 | 0.787 | 4.715 | 1.749 |
| Â | Â | BCGLM | 2596.8 | 0.790 | 4.654 | 1.737 |
| Â | Â | BCGLMM | 2388.5 | 0.814 | 4.133 | 1.637 |
500 | 0.2 | BGLM | 2282.8 | 0.720 | 3.869 | 1.583 |
| Â | Â | BCGLM | 2244.7 | 0.728 | 3.773 | 1.564 |
| Â | Â | BCGLMM | 2103.6 | 0.752 | 3.421 | 1.488 |
| Â | 0.5 | BGLM | 2839.0 | 0.778 | 5.260 | 1.849 |
| Â | Â | BCGLM | 2765.4 | 0.786 | 5.075 | 1.816 |
| Â | Â | BCGLMM | 2300.6 | 0.832 | 3.913 | 1.576 |
| Â | 0.7 | BGLM | 3213.1 | 0.788 | 6.195 | 2.009 |
| Â | Â | BCGLM | 3096.6 | 0.798 | 5.903 | 1.961 |
| Â | Â | BCGLMM | 2521.0 | 0.846 | 4.464 | 1.676 |