knitr::opts_chunk$set( collapse = TRUE, comment = "#>", fig.retina = 2, fig.path = "man/figures/README-" ) version <- as.vector(read.dcf("DESCRIPTION")[, "Version"]) version <- gsub("-", ".", version)
{readabs} helps you easily download, import, and tidy data from the Australian Bureau of Statistics within R. This saves you time manually downloading and tediously tidying data and allows you to spend more time on your analysis.
Install the latest CRAN version of {readabs} with:
install.packages("readabs")
You can install the development version of {readabs} from GitHub with:
# if you don't have devtools installed, first run: # install.packages("devtools") devtools::install_github("mattcowgill/readabs")
The ABS releases data in many different formats, through many different dissemination channels.
The {readabs} contains functions for working with three different types of ABS data:
read_abs()
and related functions downloads, imports, and tidies ABS time series data.download_abs_data_cube()
and related functions find and download ABS data cubes, which
are spreadsheets on the ABS website that are not in the standard time series format.read_api()
and related functions find, filter, and import data from the ABS.Stat API.A key function in {readabs} is read_abs()
, which downloads, imports, and tidies time series data from the ABS website. Note that read_abs()
only works with spreadsheets in the standard ABS time series format.
First we'll load {readabs} and the {tidyverse}:
devtools::load_all() library(tidyverse)
library(readabs) library(tidyverse)
Now we'll create one data frame that contains all the time series data from the Wage Price Index, catalogue number 6345.0:
all_wpi <- read_abs("6345.0")
This is what it looks like:
str(all_wpi)
It only takes you a few lines of code to make a graph from your data:
all_wpi %>% filter( series == "Percentage Change From Corresponding Quarter of Previous Year ; Australia ; Total hourly rates of pay excluding bonuses ; Private and Public ; All industries ;", !is.na(value) ) %>% ggplot(aes(x = date, y = value, col = series_type)) + geom_line() + theme_minimal() + labs(y = "Annual wage growth (per cent)")
In the example above we downloaded all the time series from a catalogue number. This will often be overkill. If you know the data you need is in a particular table, you can just get that table like this:
wpi_t1 <- read_abs("6345.0", tables = 1)
If you want multiple tables, but not the whole catalogue, that's easy too:
wpi_t1_t5 <- read_abs("6345.0", tables = c("1", "5a"))
For more examples, please see the vignette on working with time series data (run browseVignettes("readabs")
).
Some other functions that may come in handy when working with ABS time series data:
read_abs_local()
imports and tidies time series data from ABS spreadsheets stored on a local drive. Thanks to Hugh Parsonage for contributing to this functionality.separate_series()
splits the series
column of a tidied ABS time series spreadsheet into multiple columns, reducing the manual wrangling that's needed to work with the data. Thanks to David Diviny for writing this function.There are several functions that load specific ABS time series data:
read_cpi()
imports the Consumer Price Index numbers as a two-column tibble: date
and cpi
. This is useful for joining to other series to adjust data for changes in consumer prices.read_awe()
returns a long time series of Average Weekly Earnings data. read_job_mobility()
downloads, imports and tidies tables from the ABS Job Mobility dataset.The ABS (generally) releases time series data in a standard format, which allows read_abs()
to download, import and tidy it (see above). But not all ABS data is time series data - the ABS also releases data as 'data cubes'. These are all formatted in their own, unique way.
Unfortunately, because data cubes are all formatted in their own way, there is no one function that can import tidy data cubes for you in the same way that read_abs()
works with all time series. But {readabs}
still has functions that can help. Thanks to David Diviny for writing these functions.
The download_abs_data_cube()
function can download an ABS data cube for you. It works with any data cube on the ABS website. To use this function, we need two things: a catalogue_string
(the short name of the release) and cube
, a (unique fragment of) the filename within the catalogue you wish to download.
For example, let's say you wanted to download table 4 from Weekly Payroll Jobs and Wages in Australia. We can find the catalogue name like this:
search_catalogues("payroll")
Now we know that the string "weekly-payroll-jobs"
is the catalogue_string
for this release. We can now see what files are available to download from this catalogue:
show_available_files("weekly-payroll-jobs")
We want Table 4, which has the filename 6160055001_DO004.xlsx
.
We can download the file as follows:
payrolls_t4_path <- download_abs_data_cube("weekly-payroll-jobs", "004") payrolls_t4_path
The download_abs_data_cube()
function downloads the file and returns the full file path to the saved file. You can then pipe that in to another function:
payrolls_t4_path %>% readxl::read_excel( sheet = "Payroll jobs index", skip = 5 )
As it happens, if you want the ABS Weekly Payrolls data, you don't need to use download_abs_data_cube()
directly. Instead, there is a convenience function available that downloads, imports, and tidies the data for you:
read_payrolls()
There is also a convenience function available for data cube GM1 from the monthly Labour Force data, which contains labour force gross flows:
read_lfs_grossflows()
The ABS has created a new site to access its data, called the ABS Data Explorer, also known as ABS.Stat. As at early 2023, this site is in Beta mode. The site provides an API.
The {readabs} package includes functions to query the ABS.Stat API. Thank you to Kinto Behr for writing these functions. The functions are:
read_api_dataflows()
lists available dataflows (roughly equivalent to 'tables')read_api_datastructure()
lists variables within a particular dataflow and the levels of those variables, which you can use to filter the data server-side in an API queryread_api()
downloads data from the ABS.Stat API.Let's list available dataflows:
flows <- read_api_dataflows()
Say from this I am interested in the first dataflow, the projected population of
Aboriginal and Torres Strait Islander Australians. The id for this dataflow is
"ABORIGINAL_POP_PROJ"
, which I can use to download the data.
In this case, I could download the entire dataflow with:
read_api("ABORIGINAL_POP_PROJ")
Let's say I'm only interested in the population projections for males, not females or all persons. In that case, I can filter the data on the ABS server before downloading my query. I can use read_api_datastructure()
to help with this.
read_api_datastructure("ABORIGINAL_POP_PROJ")
From this, I can see that there's a variable (var
) called sex_abs
, which can take the value 1
, 2
, or 3
, corresponding to Males
, Females
and Persons
. If I only want to data for Males, I can obtain this by supplying a datakey:
read_api("ABORIGINAL_POP_PROJ", datakey = list(sex_abs = 1))
Note that in some cases, querying the API without filtering the data will return an error, as the table will be too big. In this case, you will need to supply a datakey that reduces the size of the data.
Certain corporate networks restrict your ability to download files in an R session. On some of these networks, the "wininet"
method must be used when downloading files. Users can now specify the method that will be used to download files by setting the "R_READABS_DL_METHOD"
environment variable.
For example, the following code sets the environment variable for your current session:
Sys.setenv("R_READABS_DL_METHOD" = "wininet")
You can add R_READABS_DL_METHOD = "wininet"
to your .Renviron to have this persist across sessions.
If you have other issues using {readabs}
in your corporate environment, I would appreciate you opening an issue on GitHub.
GitHub issues containing error reports or feature requests are welcome. Please try to make a reprex (a minimal, reproducible example) if possible.
Alternatively you can email the package maintainer at mattcowgill at gmail dot com.
The {readabs}
package is not associated with the Australian Bureau of Statistics.
All data is provided subject to any restrictions and licensing arrangements
noted on the ABS website.
We're pleased to be included in a list of software that can be used to work with official statistics.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.