paul-buerkner/brms

Data-related post-processing for multiple imputation models via brm_multiple()

Open

#997 opened on Sep 11, 2020

View on GitHub
 (10 comments) (0 reactions) (0 assignees)R (1,402 stars) (220 forks)batch import
featuregood first issuepost-processing

Description

When I try to use bayes_R2() on a brmsfit_multiple object, I get a warning message that only the first imputed data set was used. I assume this means that R2 was only calculated using the first imputed dataset. I asked about this on the mc-stan discourse and it seems that R2 can be pooled from the combined posterior samples just like many other parameters. Would it be possible to add that functionality to the bayes_R2() method? I tried to look into doing it myself, but was struggling to understand the fit$fit S4 object.

library(brms)
library(mice)
imp <- mice(nhanes2)
fit <- brm_multiple(bmi ~ age + hyp + chl, data = imp, chains = 1)
bayes_R2(fit)
#>    Estimate Est.Error      Q2.5    Q97.5
#> R2 0.4452458 0.1071932 0.2068827 0.623473
#> Warning message:
#> Using only the first imputed data set. Please interpret the results with caution until a more principled approach has been implemented.

Contributor guide

Tech stack
r
Domain
data
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-2 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
Familiarity with brms packageUnderstanding of multiple imputationBasic knowledge of S4 objects in R
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
30
Research direction
Examine the source code of bayes R2 in the brms package, likely located in R/bayes R2.R. Understand how it currently processes a single imputed dataset and identify the parts that need to be extended to handle multiple imputations. Look at the issue comments for discussion on pooling R2 from combined posterior samples. The implementation should loop over all imputed datasets and combine results, possibly using the fit$fit S4 object structure.