3.4 brms::fitted() results in "not an exported object from 'namespace:brms'" error about statistical_rethinking_with_brms_ggplot2_and_the_tidyverse_2_ed HOT 7 OPEN

@petzi53 the three colons are for objects that are not exported: often these are undocumented, but not necessarily.

from statistical_rethinking_with_brms_ggplot2_and_the_tidyverse_2_ed.

The error message is correct that brms::fitted() doesn't exist, at least not exactly. What exists is a function brms::fitted.brmsfit(), which is an S3 method. R then knows "when an object has the class brmsfit and the user calls fitted() on it, despatch that call to the fitted.brmsfit() method in the brms package`".

This is why when you have such an object (here that's b3.1) you can just call fitted(b3.1) and everything works, but adding the namespace explicitly doesn't. The S3 system is helping to make your life a bit easier.

from statistical_rethinking_with_brms_ggplot2_and_the_tidyverse_2_ed.

Thank you for the wonderful explication! Indeed, the b3.1 object has the class "brmsfit". But shouldn't then brms::fitted.brmsfit() work? (It generates for me the same error: 'fitted.brmsfit' is not an exported object from 'namespace:brms') This is weird for me, as the {brms} help file clearly shows that fitted.brmsfit() exists. Could it be that there is something wrong with my installation? (I add details at the end of this post.)

Using just fitted() is working, but my result does not conform with Kurtz's p values:
0.6318994, 0.8105015, 0.7677781, 0.7250286, 0.7265799, 0.7376768, 0.5400708, 0.6697464,… in contrast to Kurtz's values:
0.6878563, 0.5386513, 0.7030050, 0.6889854, 0.4738290, 0.5879676, 0.5843004, 0.7014505, …

Sorry, I still haven't read the S3 chapter in R Advanced, maybe then I would be able to solve my problem and wouldn't bother you.

Quarto check

[✓] Checking versions of quarto binary dependencies...
Pandoc version 3.1.5: OK
Dart Sass version 1.55.0: OK
Deno version 1.33.2: OK
[✓] Checking versions of quarto dependencies......OK
[✓] Checking Quarto installation......OK
Version: 1.4.232
Path: /Applications/quarto/bin

[✓] Checking basic markdown render....OK

[✓] Checking Python 3 installation....OK
Version: 3.8.3
Path: /Library/Frameworks/Python.framework/Versions/3.8/bin/python3
Jupyter: 5.3.0
Kernels: python3

[✓] Checking Jupyter engine render....OK

(-) Checking R installation...........R version 4.3.1 (2023-06-16)
[✓] Checking R installation...........OK
Version: 4.3.1
Path: /Library/Frameworks/R.framework/Resources
LibPaths:
- /Library/Frameworks/R.framework/Versions/4.3-x86_64/library
- /Library/Frameworks/R.framework/Versions/4.3-x86_64/Resources/library
knitr: 1.43
rmarkdown: 2.23

[✓] Checking Knitr engine render......OK

Session Info

R version 4.3.1 (2023-06-16)
Platform: x86_64-apple-darwin20 (64-bit)
Running under: macOS Ventura 13.4.1

Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.3-x86_64/Resources/lib/libRlapack.dylib; LAPACK version 3.11.0

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

time zone: Europe/Vienna
tzcode source: internal

attached base packages:
[1] stats graphics grDevices utils datasets methods base

other attached packages:
[1] brms_2.19.0 Rcpp_1.0.11 lubridate_1.9.2 forcats_1.0.0 stringr_1.5.0
[6] dplyr_1.1.2 purrr_1.0.1 readr_2.1.4 tidyr_1.3.0 tibble_3.2.1
[11] ggplot2_3.4.2 tidyverse_2.0.0 testthat_3.1.10 devtools_2.4.5 usethis_2.2.2

loaded via a namespace (and not attached):
[1] tensorA_0.36.2 rstudioapi_0.15.0 jsonlite_1.8.7 shape_1.4.6
[5] magrittr_2.0.3 TH.data_1.1-2 estimability_1.4.1 farver_2.1.1
[9] nloptr_2.0.3 fs_1.6.3 vctrs_0.6.3 memoise_2.0.1
[13] minqa_1.2.5 base64enc_0.1-3 htmltools_0.5.5 distributional_0.3.2
[17] curl_5.0.1 tidybayes_3.0.4 StanHeaders_2.26.27 htmlwidgets_1.6.2
[21] plyr_1.8.8 sandwich_3.0-2 emmeans_1.8.7 zoo_1.8-12
[25] cachem_1.0.8 igraph_1.5.0.1 mime_0.12 lifecycle_1.0.3
[29] pkgconfig_2.0.3 colourpicker_1.2.0 Matrix_1.6-0 R6_2.5.1
[33] fastmap_1.1.1 shiny_1.7.4.1 digest_0.6.33 colorspace_2.1-0
[37] ps_1.7.5 pkgload_1.3.2.1 crosstalk_1.2.0 projpred_2.6.0
[41] timechange_0.2.0 fansi_1.0.4 abind_1.4-5 mgcv_1.9-0
[45] compiler_4.3.1 remotes_2.4.2.1 withr_2.5.0 backports_1.4.1
[49] inline_0.3.19 shinystan_2.6.0 rethinking_2.31 gamm4_0.2-6
[53] pkgbuild_1.4.2 MASS_7.3-60 sessioninfo_1.2.2 gtools_3.9.4
[57] loo_2.6.0 tools_4.3.1 httpuv_1.6.11 threejs_0.3.3
[61] glue_1.6.2 callr_3.7.3 nlme_3.1-162 promises_1.2.0.1
[65] grid_4.3.1 cmdstanr_0.5.3 checkmate_2.2.0 reshape2_1.4.4
[69] generics_0.1.3 gtable_0.3.3 tzdb_0.4.0 hms_1.1.3
[73] xml2_1.3.5 utf8_1.2.3 pillar_1.9.0 ggdist_3.3.0
[77] markdown_1.7 posterior_1.4.1 later_1.3.1 splines_4.3.1
[81] lattice_0.21-8 survival_3.5-5 tidyselect_1.2.0 miniUI_0.1.1.1
[85] knitr_1.43 arrayhelpers_1.1-0 gridExtra_2.3 V8_4.3.3
[89] rversions_2.1.2 stats4_4.3.1 xfun_0.39 bridgesampling_1.1-2
[93] brio_1.1.3 matrixStats_1.0.0 DT_0.28 rstan_2.26.22
[97] stringi_1.7.12 boot_1.3-28.1 codetools_0.2-19 cli_3.6.1
[101] RcppParallel_5.1.7 shinythemes_1.2.0 xtable_1.8-4 munsell_0.5.0
[105] processx_3.8.2 coda_0.19-4 svUnit_1.0.6 parallel_4.3.1
[109] rstantools_2.3.1.1 ellipsis_0.3.2 prettyunits_1.1.1 dygraphs_1.1.1.6
[113] profvis_0.3.8 urlchecker_1.0.1 bayesplot_1.10.0 Brobdingnag_1.2-9
[117] lme4_1.1-34 mvtnorm_1.2-2 scales_1.2.1 xts_0.13.1
[121] crayon_1.5.2 rlang_1.1.1 multcomp_1.4-25 shinyjs_2.1.0

from statistical_rethinking_with_brms_ggplot2_and_the_tidyverse_2_ed.

Hey @petzi53, thank you for raising the issue, and thank you, @hamedbh for the helpful response. I have to confess this technical issue is brushing up agains the boundaries of my knowledge. I haven't read the S3 chapter in R Advanced either. I don't have a full solution at the moment, but I may consider dropping the language of brms::fitted() if it's misleading.

from statistical_rethinking_with_brms_ggplot2_and_the_tidyverse_2_ed.

@petzi53 if you were to try brms:::fitted.brmsfit() it should work. That’s with three colons, which allows you to access objects that aren’t exported. It’s a useful exercise to play with it and figure out the differences between ordinary functions and S3 methods, but in general I would be disinclined to directly access objects that aren’t exported in this way. The package developer provides these methods partly as a convenience to the user (because you can just call fitted(b3.1) or similar) but also to limit the user-facing parts of the package.

@ASKurz I think you’re right that removing the reference to brms::fitted() would be wise. Perhaps a whole explanation on S3 methods would be a distraction though: not sure it would really help things. Perhaps just change the text to read “… we can do so with the fitted() method”, and link to that documentation page for fitted.brmsfit() that I linked above?

from statistical_rethinking_with_brms_ggplot2_and_the_tidyverse_2_ed.

@hamedbh Yep, this worked! Thank you!

(I always thought that three colons are used for internal – not documented – functions. But fitted.brmsfit() has documentation.)

However, my results are still the same as mentioned above and very different from Kurz's version.

@ASKurz I just noticed that you have published another version. Which one is the newest?

In both versions, you mentioned a vector of 4,000 but in the example published from this repo here we both have 16,000 rows.

from statistical_rethinking_with_brms_ggplot2_and_the_tidyverse_2_ed.

@hamedbh, I have ebooks connected to the first and second edition of McElreath's textbook. This repo is for my ebook connected to his 2nd edition. The other version you linked to is for me ebook connected to his 1st edition. The content in the two is similar, but not completely overlapping.

from statistical_rethinking_with_brms_ggplot2_and_the_tidyverse_2_ed.