smooth.ES.point_lik

ES.point_lik(log=True)

Per-observation log-likelihood of the fitted model.

Mirrors R’s pointLik.adam (R/adam.R:7298): the log-density of each in-sample observation under the fitted distribution and scale, so sum(point_lik()) equals loglik. For an occurrence (intermittent) model the demand-size density is combined with the occurrence Bernoulli contribution and the zero observations carry the differential-entropy term, exactly as R does. With log=False the densities themselves are returned.

Return type:

NDArray

---

Parent Class: ES