heaven provides fast, memory-conscious tools for preparing large Danish
administrative and health registry datasets. It was developed for analysis in
the secure research environment at Statistics Denmark and uses data.table
throughout its core workflows.
The package includes tools for:
- importing selected variables and records from SAS datasets;
- splitting follow-up time into analysis intervals;
- identifying conditions from ICD, ATC, and other code lists;
- deriving prescription exposure periods;
- incidence-density and exposure-density matching;
- direct standardisation and common registry-data summaries; and
- inspecting Parquet datasets without collecting them in full.
Until the package is released on CRAN, install the development version from GitHub:
# install.packages("remotes")
remotes::install_github("tagteam/heaven")library(heaven)
library(data.table)
set.seed(7)
population <- simPop(1000)
population[]Most functions accept or return data.table objects. Because data.table can
modify objects by reference, use data.table::copy() before a call when the
original object must remain unchanged.
The SAS import helpers require access to a local or remote SAS installation.
The Parquet viewer uses the optional arrow and dplyr packages.
See vignette("R_on_DST", package = "heaven") for an introduction to working
with large datasets at Statistics Denmark. Report bugs or request features at
https://github.com/tagteam/heaven/issues.
GPL (>= 3)