RADS, APDE's 'R Automated Data System, is a suite of tools written in R and designed to make standard public health analyses faster, more standardized, and less prone to error. While some tools may be applicable for different settings, the toolset has been customized to the needs of PHSKC's APDE. While these tools have only been tested in Windows, they should work identically on a Linux or Mac OS X machine.
If you haven't yet installed rads, follow these steps:
Make sure remotes is installed ... install.packages("remotes").
RADS depends on version 17 of Microsoft ODBC Driver for SQL Server. You can download it here
Install a keyring:: 'service' for accessing HHSAW.
install.packages("keyring")
keyring::key_set(service = 'azure', username = 'Your.KCUsername@kingcounty.gov')
In the pop-up window type in your standard King County password (the one you use to log into your laptop) and click OK
Install rads ... remotes::install_github("PHSKC-APDE/rads", auth_token = NULL)
remotes::install_github("PHSKC-APDE/rads", ref = "dev", auth_token = NULL)Load rads ... library(rads)
get_population now accepts 'ccl', 'csa', 'inc_uninc', 'puma', 'kccd', and 'tribal' as valid geography types (geo_type). These refer to crisis care levy areas, community service areas, incorperated and unincorperated king county, public use microdata areas, king county council districts, and tribal lands respectively.rads::calc can now handle imputationList type inputs as data. See the [wiki](https://github.com/PHSKC-APDE/rads/wiki/calc for more information.rads::calc now has sorted outputs. This is a a minor, potentially breaking change.get_data_hys updated to pull 2023 data.get_population data updated to pull estimates by hispanic/non-hispanic within a race (as opposed to Hispanic as its own race).calcget_population, get_data*, and list_dataset_columns() functions.get_population now defaults to 2020 vintage geographies (e.g. new HRAs)get_population has been overhauled and now is able to access frankenpop aka Population Interim Estimates (PIE). For the most part, existing get_population calls should still work-- although the estimates being pulled will differ slightly:
round now defaults to FALSE. Now you can have fractional people. We recommend rounding before publishing/sharing results so as not to confuse people.
Frankenpop estimates are now the default population numbers as governed by census_vintage = 2020. Setting the census_vintage argument to 2010 will return the old OFM SADE estimates based on Census 2010. These estimates are out of date and should only be used sparingly.
Users can specify the vintage of geography they want via the geo_vintage option. geo_vintage (current default is 2010 although this will change to 2020 on the next release) refers to the decennial census associate with a class of geography. For example, all census geographies change with each decennial census and so do political districts.
Queries, especially for region and HRA data should be noticiably faster.
calc no longer requires a dtsurvey type object as an input. data.frames and other dataset types are converted on the fly based on the needs of calc. This means you dplyr heathens can do some data cleaning with your pipes and mutations without having to bounce back and forth between data types.Version 1.0.0 includes a major overhaul of the workhorse calc function and the cleaning up of some dependencies.
Potential breaking changes:
Calc will only accept objects of type dtsurvey, dtrepsurvey, or dtadmin. The documentation for these types (and the functions used to create them) can be found in the dtsurvey package.
Subsetting a dataset via calc no longer makes use of a dplyr::filter-esque interface. Instead, via the where argument, a user may pass an unquoted (e.g., no " or ') expression that would be valid in the i part of a DT[i,j,by] command.
The default method for computing confidence intervals for survey data has changed. The practical difference of this change is essentially nil, but interested users can fuss around with the differences between the xlogit (new) and logit methods to computing CIs via survey::svyciprop.
After installation, we highly recommend that you start by walking through a vignette on the wiki. * calc() * get_population() * get_data() * age_standardize()