The cancensus package was developed to provide users with a way to access Canadian Census in a programmatic way following good tidy data practices. While the structure and data in cancensus is unique to Canadian Census data, this package is inspired in part by tidycensus, a package to interface with the US Census Bureau data APIs.
As Statistics Canada does not provide direct API access to Census data, cancensus retrieves Census data indirectly through the CensusMapper API. CensusMapper is a project by Jens von Bergmann, one of the authors of cancensus, to provide interactive geographic visualizations of Canadian Census data. CensusMapper databases store all publicly available data from Statistics Canada for the 2006, 2011, and 2016 Censuses. Censusmapper data can be accessed via an API and cancensus is built to interface directly with it.
cancensus requires a valid CensusMapper API key to use. You can obtain a free API key by signing up for a CensusMapper account. CensusMapper API keys are free and public API quotas are generous; however, due to incremental costs of serving large quantities of data, there limits to API usage in place. For most use cases, these API limits should not be an issue. Production uses with large extracts of fine grained geographies may run into API quota limits. For larger quotas, please get in touch with Jens directly.
To check your API key, just go to “Edit Profile” (in the top-right of
the CensusMapper menu bar). Once you have your key, you can store it in
your system environment so it is automatically used in API calls. To do
so just enter
set_api_key(<your_api_key>, install = TRUE)
The stable version of cancensus can be easily installed from CRAN.
install.packages("cancensus") library(cancensus) options(cancensus.api_key = "your_api_key") options(cancensus.cache_path = "custom cache path")
Alternatively, the latest development version can be installed from
# install.packages("devtools") ::install_github("mountainmath/cancensus") devtools library(cancensus) options(cancensus.api_key = "your_api_key") options(cancensus.cache_path = "custom cache path")
For performance reasons, and to avoid unnecessarily drawing down API
quotas, cancensus caches data queries under the hood.
By default, cancensus caches in R’s temporary
directory, but this cache is not persistent across sessions. In order to
speed up performance, reduce quota usage, and reduce the need for
unnecessary network calls, we recommend assigning a persistent local
set_cache_path(<local cache path>, install = TRUE),
this enables more efficient loading and reuse of downloaded data.. Users
will be prompted with a suggestion to change their default cache
location when making API calls if one has not been set yet.
cancensus provides three different functions for
retrieving Census data: *
get_census to retrieve Census
data and geography as a spatial dataset *
to retrieve Census data only as a flat data frame *
get_census_geometry to retrieve Census geography only as a
collection of spatial polygons.
get_census takes as inputs a dataset parameter, a list
of specified regions, a vector of Census variables, and a Census
geography level. You can specify one of three options for spatial
NA to return data only,
sf to return
an sf-class data frame, or
sp to return a
# Returns a data frame with data only <- get_census(dataset='CA21', regions=list(CMA="59933"), census_data vectors=c("v_CA21_434","v_CA21_435","v_CA21_440"), level='CSD', use_cache = FALSE, geo_format = NA, quiet = TRUE) # Returns data and geography as an sf-class data frame <- get_census(dataset='CA21', regions=list(CMA="59933"), census_data vectors=c("v_CA21_434","v_CA21_435","v_CA21_440"), level='CSD', use_cache = FALSE, geo_format = 'sf', quiet = TRUE) # Returns a SpatialPolygonsDataFrame object with data and geography <- get_census(dataset='CA21', regions=list(CMA="59933"), census_data vectors=c("v_CA21_434","v_CA21_435","v_CA21_440"), level='CSD', use_cache = FALSE, geo_format = 'sp', quiet = TRUE)
cancensus utilizes caching to increase speed,
minimize API token usage, and to make data available offline. Downloaded
data is hashed and stored locally so if a call is made to access the
same data, cancensus will read the local version
instead. To force cancensus to refresh the data,
use_cache = FALSE as a parameter for
Additional parameters for advanced options can be viewed by running
cancensus can access Statistics Canada Census data
for Census years 1996, 2001, 2006, 2011, and 2016 . You can run
list_census_datasets to check what datasets are currently
available for access through the CensusMapper API. Additional data for
the 2016 Census will be included in Censusmapper within a day or two
after public release by Statistics Canada. Statistics Canada maintains a
release schedule for the Census 2016 Program which can be viewed on
Thanks to contributions by the Canada Mortgage and Housing Corporation (CMHC), cancensus now includes additional Census-linked datasets as open-data releases. These include annual taxfiler data at the census tract level for tax years 2000 through 2017, which includes data on incomes and demographics, as well as specialized crosstabs for Structural type of dwelling by Document type, which details occupancy status for residences. These crosstabs are available for the 2001, 2006, 2011, and 2016 Census years at all levels starting with census tract.
list_census_datasets() will show all
available datasets alongside their metadata.
As other Census datasets become available via the CensusMapper API,
they will be listed as output when calling
Census data is aggregated at multiple geographic levels. Census geographies at the national (C), provincial (PR), census metropolitan area (CMA), census agglomeration (CA), census division (CD), and census subdivision (CSD) are defined as named census regions.
Canadian Census geography can change in between Census periods.
cancensus provides a function,
list_census_regions(dataset), to display all named census
regions and their corresponding id for a given census dataset.
regions parameter in
requires as input a list of region id strings that correspond to that
regions geoid. You can combine different regions together into region
# Retrieves Vancouver and Toronto list_census_regions('CA21') %>% filter(level == "CMA", name %in% c("Vancouver","Toronto")) <- get_census(dataset='CA21', regions=list(CMA=c("59933","35535")), census_data vectors=c("v_CA21_434","v_CA21_435","v_CA21_440"), level='CSD', use_cache = FALSE, quiet = TRUE)
Census data accessible through cancensus comes is available in a number of different aggregation levels including:
|Code||Description||Count in Census 2016|
|CMA||Census Metropolitan Area||35|
|EA||Enumeration Area (1996 only)||-|
|DB||Dissemination Block (2001-2016)||489676|
|Regions||Named Census Region|
regions = "59933" and
level = "CT" will return data for all 478 census tracts in
the Vancouver Census Metropolitan Area. Selecting
level = "DA" will return data for all 3450 dissemination
areas and selecting
level = "DB" will retrieve data for
15,197 dissemination block. Working with CT, DA, EA, and especially DB
level data significantly increases the size of data downloads and API
usage. To help minimize additional overhead, cancensus
supports local data caching to reduce usage and load times.
level = "Regions" will produce data strictly for
the selected region without any tiling of data at lower census
aggregation levels levels.
Census data contains thousands of different geographic regions as well as thousands of unique variables. In addition to enabling programmatic and reproducible access to Census data, cancensus has a number of tools to help users find the data they are looking for.
list_census_vectors(dataset) to view all available
Census variables for a given dataset.
For each variable (vector) in that Census dataset, this shows:
Each Census dataset features numerous variables making it a bit of a
challenge to find the exact variable you are looking for. There is a
find_census_vectors(), for searching through
Census variable metadata in a few different ways. There are three types
of searches possible using this function: exact search, which simply
looks for exact string matches for a given query against the vector
dataset; keyword search, which breaks vector metadata into unigram
tokens and then tries to find the vectors with the greatest number of
unique matches; and, semantic search which works better with search
phrases and has tolerance for inexact searches. Switching between search
modes is done using the
query_type argument when calling
# Find the variable indicating the number of people of Austrian ethnic origin find_census_vectors("Australia", dataset = "CA16", type = "total", query_type = "exact") find_census_vectors("Australia origin", dataset = "CA16", type = "total", query_type = "semantic") find_census_vectors("Australian ethnic", dataset = "CA16", type = "total", query_type = "keyword", interactive = FALSE)
Census variables are frequently hierarchical. As an example, consider
the variable for the number of people of Austrian ethnic background. We
can select that vector and quickly look up its entire hierarchy using
parent_census_vectors on a vector list.
list_census_vectors("CA16") %>% filter(vector == "v_CA16_4092") %>% parent_census_vectors()
Sometimes we want to traverse the hierarchy in the opposite direction. This is frequently required when looking to compare different variable stems that share the same aggregate variable. As an example, if we want to look the total count of Northern European ethnic origin respondents disaggregated by individual countries, it is pretty easy to do so.
# Find the variable indicating the Northern European aggregate find_census_vectors("Northern European", dataset = "CA16", type = "Total")
The search result shows that the vector v_CA16_4092
represents the aggregate for all Northern European origins. The
child_census_vectors function can return a list of its
constituent underlying variables.
# Show all child variable leaves list_census_vectors("CA16") %>% filter(vector == "v_CA16_4122") %>% child_census_vectors(leaves = TRUE)
leaves = TRUE parameter specifies whether
intermediate aggregates are included or not. If
only the lowest level variables are returns - the “leaves” of the