We identified 22 R packages on CRAN that deal with US Census data. Since the Guide is intended to be a brief introduction to the field, we felt that simply listing these packages in alphabetical order, along with a brief description, might not provide the best value to readers. In particular, we were concerned that it might overwhelm readers and not provide sufficient guidance.
In the end we decided to:
- Include a list of the five most popular packages in the main text of the Guide. Sort this list by the number of downloads each package had in the last year.
- Include additional packages which have important functionality that is not present in the original list.
- Default to describing these packages as CRAN does. Allow package authors to modify these descriptions.
- List all 22 packages in a separate page.
We used the following script to determine the number of downloads each package had in the last year:
library(cranlogs)
library(tidyverse)
# 33 packages that, based on their description, appear related to US
# census data
packages = c("acs",
"cdltools",
"censable",
"census",
"censusapi",
"censusGeography",
"censusr",
"choroplethr",
"choroplethrMaps",
"cpsR",
"cpsvote",
"easycensus",
"idbr",
"ipumsr",
"noncensus",
"ppmf",
"predictrace",
"SeerMapper",
"SeerMapper2010East",
"SeerMapper2010Regs",
"SeerMapper2010West",
"SeerMapperEast",
"SeerMapperRegs",
"SeerMapperWest",
"tidycensus",
"tigris",
"totalcensus",
"UScensus2000cdp",
"UScensus2000tract",
"UScensus2010",
"usdata",
"us.census.geoheader",
"zctaCrosswalk")
# use last year's data
df = cran_downloads(packages=packages, from="2017-10-24", to="2024-05-31")
# sort by package
df2 = df %>%
group_by(package) %>%
summarize(annual_downloads = sum(count))
total_downloads = sum(df2$annual_downloads)
df2$percent = round(df2$annual_downloads / total_downloads * 100)
df2 %>%
arrange(-annual_downloads) %>%
print(n = Inf)
The script gives us the following output:
# package annual_downloads percent
# <chr> <dbl> <dbl>
# 1 tigris 727026 24
# 2 tidycensus 526920 18
# 3 usdata 265676 9
# 4 choroplethr 199049 7
# 5 acs 192425 6
# 6 ipumsr 190174 6
# 7 censusapi 139707 5
# 8 choroplethrMaps 133883 4
# 9 noncensus 99437 3
# 10 totalcensus 73907 2
# 11 UScensus2010 36945 1
# 12 idbr 36860 1
# 13 censusr 32909 1
# 14 UScensus2000tract 27003 1
# 15 censusGeography 26732 1
# 16 UScensus2000cdp 25146 1
# 17 SeerMapper 24432 1
# 18 SeerMapper2010Regs 24308 1
# 19 SeerMapper2010West 24266 1
# 20 SeerMapperWest 24207 1
# 21 SeerMapper2010East 24093 1
# 22 SeerMapperEast 23988 1
# 23 SeerMapperRegs 23897 1
# 24 predictrace 21903 1
# 25 censable 12691 0
# 26 cpsvote 11287 0
# 27 cpsR 10301 0
# 28 us.census.geoheader 9546 0
# 29 easycensus 7278 0
# 30 ppmf 6227 0
# 31 census 5047 0
# 32 zctaCrosswalk 2552 0
# 33 cdltools 0 0
This data shows that some packages in our list are much more popular than others. Additionally, popularity decreasing sharply.
At this point we considered listing only the five most popular packages in the main page of the Guide. This would mirror how the Guide treats the 100+ programs that Census conducts. There we include only the five most popular programs in the main body of the text. But we also include a link to a separate page that lists all of the programs.
The problem with only looking at package popularity is that some of the less popular packages include important functionality. In particular, we felt it was important to include the censusapi and ipumsr packages in the main body of the text because:
censusapiaims to provide access to the entirety of the Census Bureau's API. We felt that awareness of this package is important when discussing how R's package ecosystem facilitates working with the US Census Bureau. While most of the packages in the list already provided access to some of the Census Bureau's API, none of them provide access to all of it.ipumsrfacilitates working with data from IPUMS (Integrated Public Use Microdata Series). IPUMS is the world's largest individual-level population database, and is an important resource in the US Census ecosystem. None of the packages in our original list provided this functionality.
In the original draft of this Guide we copied the package descriptions which appear on CRAN verbatim. However, we were concerned that those descriptions might be too sparse. So we invited the package authors to modify them.
As an example of this difference, here is how the ipumsr package is
described on CRAN:
ipumsr: Read 'IPUMS' Extract Files
An easy way to import census, survey and geographic data provided by 'IPUMS' into R plus tools to help use the associated metadata to make analysis easier. 'IPUMS' data describing 1.4 billion individuals drawn from over 750 censuses and surveys is available free of charge from our website https://ipums.org.
And here is how package is described in the Guide:
ipumsr. The ipumsr package assists in the import and use of IPUMS extracts in R. IPUMS data is a repackaging of census and survey data from around the world to a format harmonized across space and time and with enhanced documentation. With IPUMS data, it easy to study change, conduct comparative research, merge information across data types, and analyze individuals within family and community context. Users can go to the IPUMS website and create an extract with only the samples and variables they are interested in, and then import their downloaded extract with the ipumsr package. Work has begun on an API for IPUMS data, but it is not known when it will be available. Several IPUMS projects include data from the US Census Bureau:
- IPUMS USA repackages the public use microdata from the US Census and American Community Survey from 1850 to the present.
- NHGIS provides population, housing, agricultural, and economic data, along with GIS-compatible boundary files, for geographic units in the United States from 1790 to the present.
- IPUMS CPS provides the Current Population Survey which is a monthly workforce survey conducted by the Census Bureau.
- IPUMS International includes the US Census microdata in the same format as census data from 93 other countries.