NEWS.md
zip2town crosswalk, based on 2020 / 2022 geographies. The columns included in the data frame are slightly changed.qwi_industry: the API now uses COGs for Connecticut instead of counties.Bumping package versions just to draw attention to the fact that there’s now a set of PUMA proxy crosswalks; see proxy_pumas.
Edit xwalk—there were still more FIPS codes to update with their COG-based versions. The data frame now includes COG-based codes for block groups, tracts, towns, and PUMAs.
MINOR BREAKING CHANGE: This update corresponds to the 2022 ACS data release, which is the first to use COGs instead of counties. Because COGs have different FIPS codes, town and tract FIPS codes (but apparently not block groups) have changed to match. The bulk of their code digits stay the same, but the portion signifying the county changed, e.g. 09009140101 is now 09170140101. To deal with that without breaking too much code, there are a few changes to the package:
bridgeport_tracts, etc) have the previous county-based FIPS codes in the column geoid, and the new COG-based FIPS codes in the column geoid_cog.xwalk now has columns for COG-based town and tract FIPS codes, in addition to the previous county-based ones.multi_geo_acs with counties = "all" (the default) will get you COGs, but multi_geo_decennial will get you counties, because the switch was not retroactive.multi_geo_acs and used for names in the regions list are the ones the Census Bureau uses. Unfortunately, these aren’t all the ones the state uses. For that, I’ve added a function fix_cogs, which replaces common names for them with the ones the state lists somewhat officially, e.g. Capitol COG is in the census data, Capitol Region COG is what the state usually uses but probably not always.multi_geo_* functions took neighborhood names, weights, and GEOIDs as bare column names, with defaults (name, weight, and geoid, respectively). These now have to all be given as strings (i.e. in quotation marks), and geoid no longer has a default. This is to deal with the fact that some calculations will now need the neighborhood lookup tables’ geoid columns, and some will need geoid_cog. This only matters when you’re including neighborhoods in function calls.village2town is based on. They now overlap with towns even less well than they used to. The table has been recalculated, with towns and villages joined based on overlapping population from the 2020 decennial, and now includes populations and weights in the crosswalk. That means things could break if you’re expecting one set of CDPs and get another, or if you’re not expecting new columns in that table.multi_geo_decennial now defaults to 2020. Because the 2020 decennial uses a different summary file code from previous years, the default sumfile argument, if used with 2010, will lead to an error.
decennial_vars20. The 2010 ones are still in decennial_vars10.cb_avail, has the years, programs (ACS vs decennial), and dataset codes (SF1, ACS5, DHC, etc.) available from the Census Bureau’s API.dh_scaffold was poorly named and not a great fit for the aims of this project. It’s been moved to {stylehaven}; find it there as scaffold_project.multi_geo_acs, adj_inflation base year, label_acs.acs_vars20 with acs_vars21.multi_geo_decennial now takes "pl" as a possible value for summary file, since the full 2020 Decennial data still aren’t out.regions list. Connecticut adopted these in 2022 to replace counties. Definitions from CTOPM here.rescale option to sub_nonanswers—its default won’t change any existing codeBugfix: occupational codes have larger groups and smaller groups. One larger group (Healthcare Practitioners and Technical Occupations) was mislabeled so it was marked as being under Education, Legal, Community Service, Arts, and Media Occupations.
Some updates to 2020
acs_vars19 has been replaced by acs_vars20, and multi_geo_acs now uses 2020 as the default. Some examples & vignette code have been updated to match.Major exciting overhaul! This was the first time I felt like enough of this package is flexible and well thought out to consider it a real release. A lot of the changes are under the hood–I split a lot of functions into slimmed-down main “caller” functions and multiple task-focused “helper” functions, making it easier to maintain the package, add or modify features, and use the same code for multiple tasks.
cli for cleaner and clearer messaging (printouts on what fetch functions are getting, limitations to function calls, etc)qwi_industry and laus_trend.multi_geo_* functions–I don’t really remember why I made these plural, but they’re now singular. So a column that would have been e.g. “1_state”, “2_counties”, “3_towns” will now be “1_state”, “2_county”, “3_town”. This might break filtering you’ve done by level.acs_quick_map –> quick_map
camiller::calc_shares
gnh_tenure
Since the 2020 ACS is delayed, I decided we should still have copies of 2019 geography-related files. This should be temporary, but for now there are 2 versions of the tract shapefile (tract_sf and tract_sf19), and 2 versions of each neighborhood-tract weight table (e.g. new_haven_tracts and new_haven_tracts19, and so on). Once all the data is out, I’ll remove the 2019 versions and bump up the package version.
tract_sf and town_sf to 2020 boundaries. Don’t expect anything should have changed for towns, but many tracts were added after the Census Bureau released redistricting data.nhv_tracts to new_haven_tracts to match those for other cities.separate_acs as a very lazy way to split ACS labels.read_xtab, read_weights, xtab2df, and collapse_n_wt for working with DataHaven Community Wellbeing Survey crosstabs—see vignettesadd_logo with built-in DataHaven logosub_nonanswers, xwalk
multi_geo_acs & multi_geo_decennial call janitor::clean_names before returning. This keeps columns aligned properly if neighborhoods are included.NEWS.md file to track changes to the package.adj_inflation.multi_geo_acs and multi_geo_decennial can aggregate neighborhood data. There’s an example in the workflow vignette.dplyr 1.0.0 & tidyr 1.0.0 functions.jenks, dh_scaffold