Table 2. Prediction accuracy of different scoring methods on synthetic datasets.
BDe | AIC | BIC | fNML | |
D1 | 0.945 | 0.964 | 0.909 | 1.000 |
D2 | 0.982 | 1.000 | 0.927 | 0.964 |
D3 | 0.982 | 1.000 | 0.945 | 1.000 |
D4 | 0.964 | 0.982 | 0.982 | 1.000 |
D5 | 0.945 | 1.000 | 0.891 | 1.000 |
D6 | 0.945 | 0.982 | 0.982 | 0.964 |
D7 | 0.982 | 0.982 | 0.927 | 0.982 |
D8 | 1.000 | 0.982 | 0.964 | 0.982 |
D9 | 0.982 | 0.982 | 0.927 | 0.982 |
D10 | 0.891 | 0.945 | 0.945 | 0.964 |
Avg | 0.962 | 0.982 | 0.940 | 0.984 |
SD | 0.031 | 0.017 | 0.030 | 0.016 |
BDe: Bayesian Dirichlet Equivalent; AIC: Akaike Information Criterion; BIC: Bayesian Information Criterion; fNML: factorized Normalized Maximum Likelihood.