Table 5 Model comparison using LOOIC and expected log predictive density.

From: Modeling invasion risk of Mimosa pigra L. in Northeastern Thailand using Bayesian count models

Model

LOOIC

SE

elpd_loo

SE

p_loo

SE

elpd_diff (SE)

Negative Binomial

311.0

14.3

−155.5

7.1

21.5

4.1

0.0 (0.0)

Poisson

311.9

16.9

−155.9

8.5

24.9

4.8

−0.4 (1.6)

  1. Note. LOOIC (Leave-One-Out Information Criterion) indicates model predictive performance, with lower values reflecting better fit. elpd_loo is the expected log pointwise predictive density computed via leave-one-out cross-validation. p_loo denotes the effective number of parameters. elpd_diff reflects the difference in elpd_loo relative to the best-performing model.
Back to article page