## models.elpd_waic()


WAIC from pointwise log-likelihood draws: [lppd](models.lppd.md#prospicio.models.lppd) less the variance of


Usage


``` python
models.elpd_waic(log_lik)
```


each observation's log-likelihood.


## Parameters


`log_lik: list of list of float`  
One row per posterior draw, one column per observation.


## Returns


`Elpd`