Changes to prepare for release on CRAN. Released May 4, 2021.
Released January 19, 2021.
Removed the dependency on the
usmap package, which was removed from CRAN on January 10, 2021.
plot.covidcast_signal() now supports mapping Puerto Rico in bubble plots.
state_geo built-in datasets have been removed. These were used to support bubble plots, but the changes to
plot.covidcast_signal() rendered them unnecessary.
name_to_cbsa(), are now much faster when given input vectors with many duplicate values.
Released January 13, 2021.
plot.covidcast_signal() method now supports making choropleth plots for metropolitan areas and hospital referral regions.
as.covidcast_signal() generic function makes it easy to turn any data frame, such as data loaded from an external source, into a
covidcast_signal object that can be plotted and wrangled using the functions in this package. See
vignette("external-data") for examples.
covidcast_meta() now support signals with
time_type = "week". Select signals in the API are offered at weekly resolution. They also now support signals with
geo_types of “nation” and “hhs”, corresponding to national estimates and Department of Health & Human Services Regional Offices (“HHS regions”).
covidcast_signal() now batches requests, so that many days of data can be fetched in one API call. This dramatically improves the speed of fetching state-, MSA-, and HRR-level data, since many days of data can be fetched in one API call. County-level signals, such as cases and deaths, may still require one API call per day, since the API’s row limit is only slightly larger than the number of counties in the United States.
covidcast_signal() now fetches data from the API server in CSV format, rather than JSON, which requires less bandwidth and parsing.
covidcast_cor() is dramatically faster than before.
Released October 31, 2020.
covidcast_cor()function now interprets its
dt_yarguments in the opposite way as before, i.e. what was once a positive lag is now a negative lag. See the documentation for details.
covidcast_signals()function allows users to query multiple signals from the API at once, returning a list of data frames. The
aggregate_signals()function can convert this list into a “wide” or “long” format data frame for further analysis, while
covidcast_longer()can convert from “long” to “wide” format and vice versa.
Released August 22, 2020.
New utility functions
cbsa_to_name() to find the names corresponding to FIPS codes (for counties) or CBSA codes (for metropolitan statistical areas).
Released July 26, 2020.
Name change: The package is now called covidcast, not covidcastR. If already installed, users should remove covidcastR using
remove.packages() and then install covidcast using the instructions in
Incompatible change: The
covidcast_signal() function now takes
end_day arguments as strings in the form YYYY-MM-DD, rather than YYYYMMDD, for consistency with
as.Date() and other common R usage. The arguments can also be passed as
Date objects directly.
New plotting and mapping features provide convenient visualizations of signals. See
vignette("plotting-signals") for examples.
New feature: The
covidcast_signal() function now supports arguments
lag, allowing users to request specific versions of data in the API. This allows users to track revisions of data and to see only data that was available as of a specific date. See
vignette("covidcast") for examples.
covidcast_signal()is now a data frame augmented with extra attributes indicating the geography type and signal metadata. Additional
signalcolumns are provided, in anticipation of options to request more than one signal at once. Methods for
summaryare provided (as
summary.covidcast_signal()) to conveniently view these objects. They can still be used as ordinary data frames as well.