In nickseedorff/bmrarm: Bayesian Mixed Response Autoregressive Models

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(bmrarm)

Replicate Paper Simulations (not run)

The actual simulations took a non-trival amount of time and resulted in large objects. The following code can be used, along with the generated datasets that are stored in the package, to refit those exact models.

AR Data

Data is generated using the definition of bmrarm with $\rho=0.35$

Fit Models

Calculate cDIC and mDIC

Assess convergence of bmrarm

Forecast quality

Slope Data (no AR term)

Data is generated using the definition of bmrarm with $\rho=0$. This can be seen as a bivariate hierachical model with subject specific slopes and intercepts.

Fit Models

Calculate cDIC and mDIC

Minimal Working Example

Since the simulation studies are not actually replicated in the vignette due to time constraints, we show a minimal working example. This uses the first 5 datasets, but obtains a much smaller number of draws.

AR Data

Data is generated using the definition of bmrarm with $\rho=0.35$

Fit Models

Calculate cDIC and mDIC

Assess convergence of bmrarm

Forecast quality

nickseedorff/bmrarm documentation built on April 17, 2025, 9:43 p.m.