Table 2. Comparison of fitness of different parametric frailty models based on the Akaike information criteria, June 2018.
Baseline distribution |
Frailty distribution | Log- likelihood | K | c | AIC Value |
---|---|---|---|---|---|
Exponential | Gamma | -1404 | 10 | 1 | 2829 |
Weibull | Gamma | -603 | 10 | 2 | 1229 |
Log-logistic | Gamma | -1343 | 10 | 2 | 2711 |
Log-normal | Gamma | -1334 | 10 | 2 | 2691 |
AIC = -2lnL + 2(k +c), where c is the number of model-specific distributional parameters & k is the number of model covariates.