Addition of AR(1) process in brms model results in high pareto-k values from loo

[t-davidson]

Linking to some other relevant posts:

This post discusses potential issues with AR(1) and negative binomial: Brms: fitting negative binomial model with an AR-correlation structure - #2 by martinmodrak

This post discusses issues of high pareto-k values when estimating hierarchical negative binomial models: Hierarchical negative binomial - pareto k > 0.5