-
@_inundata talking about research compendiums for preserving analysis. RStudio Project/Package structure can be basis of this. #rstudioconfPermalink On twitter.com
Mention
❤️ 2 Favorites Mood 0
-
@_inundata Compendia can be simple to complex: medium ones might use NAMESPACE, custom functions, complex ones may have workflows and Dockerfiles as wellPermalink On twitter.com
-
@_inundata Data management: small tabular data can be stored in compendium. piggyback package can attach large data files (up to 2 gb) to github repo releases. arkdb allows you to chunk data in and out of db backends. #rstudioconfPermalink On twitter.com
-
@_inundata Isolate computing environments - package versions can result in different analyses! Think about dockerfiles - use github.com/o2r/containeritPermalink On twitter.com
-
@_inundata mybinder.org lets you share analyses that are in notebooks - can rerun analyses #rstudioconfPermalink On twitter.com
-
@_inundata Setting up binder - add text file called runtime.txt - rocker will pull appropriate container, install.R to specify packages. #rstudioconfPermalink On twitter.com
-
Think about using drake @rOpenSci package for workflows #RstudioconfPermalink On twitter.com
-
Think about long term archiving of compendia using zenodo #rstudioconfPermalink On twitter.com
-
More about compendia here github.com/karthik/rstudio2019On twitter.com