Table 1. Parameter estimates, model goodness of fit, model complexity, and predictive accuracy (case study I).
| Whole data analysis | 200 CVs | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Model | Predictors | Log likelihoodb | Effective number of parameters (pD) | Deviance information criteria (DIC) | Average CV-AUCc | Proportion of times (of 200 CVs) model in column had AUC > model in row | |||||||||||
| Age at diagnosis | Racea | Lobular (Y/N) | Tumor subtype | Pathological stage | Gene expression | M2 | M3 | M4 | M5 | M6 | M7 (COV) | M8 (COV + WGGE) | |||||
| M1 | X | −146.1 | 2.1 | 294.3 | 0.557d (0.007) | 0.14 | <0.01 | >0.99 | >0.99 | >0.99 | >0.99 | >0.99 | |||||
| M2 | X | −147.5 | 2.0 | 296.9 | 0.525d,e (0.023) | 0.59 | >0.99 | >0.99 | >0.99 | >0.99 | >0.99 | ||||||
| M3 | X | −144.3 | 2.0 | 290.6 | 0.526e (0.020) | >0.99 | >0.99 | >0.99 | >0.99 | >0.99 | |||||||
| M4 | X | −138.6 | 4.1 | 281.3 | 0.618f (0.013) | 0.14 | >0.99 | >0.99 | >0.99 | ||||||||
| M5 | X | −142.4 | 2.0 | 286.9 | 0.596f (0.012) | >0.99 | >0.99 | >0.99 | |||||||||
| M6 | X | −132.4 | 15.5 | 280.3 | 0.659g (0.011) | >0.99 | >0.99 | ||||||||||
| M7: COV | X | X | X | X | X | −146.3 | 3.2 | 295.8 | 0.704h (0.007) | >0.99 | |||||||
| M8: COV + WGGE | X | X | X | X | X | X | −131.3 | 17.6 | 280.3 | 0.721i (0.010) | |||||||
a
African American, Y/N.
b
Estimated posterior mean of the log likelihood.
c
Average over 200 tenfold CVs.
d,e,f,g,h,i
The same letter indicates that the models are no different (empirical P < 0.05).