Uniform | Poisson | Binomial | Negative Binomial | Beta-Binomial | Empirical | BMA | |
---|---|---|---|---|---|---|---|
# of Parameters | 1 | 1 | 2 | 2 | 2 | 12 | 1.4 |
BIC | 67 | 69 | 72 | 68 | 69 | 83 | 66 |
Weights | 41% | 16% | 4% | 25% | 14% | 0% | - |
The Bayesian Average Model is a notable improvement over the individual approaches.
Cross-Validated BMA with 12 folds provides a minor improvement:
We can visualize the model uncertainty estimated by cross-validation: