In smorsink1/ncov2019: Import and Visualize COVID-19, Zika, and SARS Data

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(ncov2019)

Importing Data

To easily import data on COVID-19, Zika, or SARS, use the corresponding import function:

covid_data <- importCovidData()    # default from_web = TRUE
zika_data <- importZikaData()      # default from_web = FALSE
sars_data <- importSARSData()      # default from_web = FALSE

Because Zika data and SARS data are no longer changing, it is not necessary to load data from the web. By setting from_web = FALSE, the function simply loads the data from the data directory. For COVID-19, to get the most up-to-date information, we need from_web = TRUE (the default). If you set from_web = FALSE for importCovidData(), the function will load the saved COVID-19 data from the data directory, which only goes through 2020-03-06.

Behind the scenes, the import functions are using other functions for data acquisition and cleaning. These functions are also exported for the user to observe and potentially modify. importCovidData() calls scrapeCovidData() to pull data from the web, then tidyCovidData() to reformat it, then accumulateCovidData() to pull together data on cases, deaths, and recovered. importZikaData() calls scrapeZikaData() to pull data from the web, then reformatZikaData() to reformat it, then cleanZikaData() to reconcile data reporting differences. importSARSData() calls scrapeSARSData() to pull data from the web, then cleanSARSData() to reformat it and address data issues.

Filtering Data

After importing disease data, the next step is to filter the data. filterDiseaseData() allows for easy filtering of the data by date, region (country), province (more specific location information than country), value type ("cases", "deaths", or "recovered"), and value (number of cases, deaths, or recovered). Although other data filtering options exist, filterDiseaseData() is preferrable in the context of the ncov2019 package. The output is guaranteed to be compatible with the plotting functions, and filterDiseaseData() checks the input data for compatibility as well.

us_covid_data <- filterDiseaseData(covid_data, country = "US")
head(us_covid_data)

us_covid_cases_above_100 <- filterDiseaseData(covid_data, country = "US", type = "cases", min_value = 100)
head(us_covid_cases_above_100)

us_italy_covid_data <- filterDiseaseData(covid_data, country = c("US", "Italy"))
head(us_italy_covid_data)

china_covid_cases_deaths <- filterDiseaseData(covid_data, country = "China", type = c("cases", "deaths"))
head(china_covid_cases_deaths)

covid_data_march_20 <- filterDiseaseData(covid_data, first_date = "2020-03-20", last_date = "2020-03-20")

zika_data_may <- filterDiseaseData(zika_data, first_date = "2016-05-01", last_date = "2016-05-31")
head(zika_data_may)

zika_data_may_include_suspected <- filterDiseaseData(zika_data, first_date = "2016-05-01", last_date = "2016-05-31", include_suspected = TRUE)
head(zika_data_may_include_suspected)

sars_deaths_june <- filterDiseaseData(sars_data, type = "deaths", first_date = "2003-06-01", last_date = "2003-06-30")
head(sars_deaths_june)

Plotting Data

After filtering the data down to the data set that you want to visualize, use the plotting functions. mapPlotStatic() and mapPlotAnimate() display data on a world map.

mapPlotStatic() displays one day of data: use the selected_date argument to specify which day, and use the selected_value_type to specify if you want to visualize "cases", "deaths", or "recovered". Note the use of the congregate argument to search for cumulative data during the week near selected_date. This is necessary when the data does not have any reporting on selected_date. The Zika data does not have reporting every day, thus congregate = TRUE is recommended when using mapPlotStatic() on Zika data.

mapPlotStatic(covid_data, selected_date = "2020-03-20", selected_value_type = "cases")
# mapPlotStatic(covid_data, selected_date = "2020-03-20", selected_value_type = "deaths")
# mapPlotStatic(covid_data, selected_date = "2020-03-20", selected_value_type = "recovered")
# mapPlotStatic(us_covid_cases_above_100, selected_date = "2020-03-20")
### if the data only includes one value_type, you don't need to specify selected_value_type
# mapPlotStatic(covid_data_march_20, selected_value_type = "cases")
### if the data only includes one date, you don't need to specify selected_date

# for some of the data data, data was reported on different days
# so, the cumulative numbers for 2016-05-28
mapPlotStatic(zika_data_may_include_suspected, selected_date = "2016-05-28")
# don't quite match the cumulative numbers if we look at the week around 2016-05-28
mapPlotStatic(zika_data_may_include_suspected, selected_date = "2016-05-28", congregate = TRUE)
# hence, congregate = TRUE is recommended for mapPlotStatic on zika data

### sometimes, mapPlotStatic won't plot anything if there was no reporting on that date
# mapPlotStatic(zika_data_may_include_suspected, selected_date = "2016-05-29")
### in these cases, congregate = TRUE is required
# mapPlotStatic(zika_data_may_include_suspected, selected_date = "2016-05-29", congregate = TRUE)

mapPlotAnimate() displays multiple days of data in an animation. The first_date and last_date arguments can be used to specify the starting and ending dates, the selected_value_type specifies whether you want to visualize "cases", "deaths", or "recovered". The animation will be displayed in the "Viewer" window in RStudio.

Note: Below, warnings have been suppressed on the mapPlotAnimate() function. The reason for this is that since the sizing of the points on the animation plot is log10(value), all rows which are value = 0 will output a size of negative infinity, since log10(0) = -Inf. Attempting to plot points of size -Inf is what returns a warning. However, these rows of value = 0 must be kept in the animated data frame in order to create a fluid animation using the gganimate package. Additionally, since value = 0, we don't want these rows plotted anyways. Thus, these warnings remain, but they do not deter the construction of the animation in any way.

suppressWarnings(
  mapPlotAnimate(us_covid_data, first_date = "2020-03-01", last_date = "2020-03-15")
)

plotTimeSeries() displays various disease statistics over time. Specify the y-axis variable with the plot_what argument. See documentation for all options for plot_what. Specify the layers of the plot with the group argument. For example, group = "region" will create one layer of the plot for each region in the data. The x-axis variable plots time, either "date" or "day_of_disease" (number of days since the number of cases in the group reaches 100). Specify the x_axis variable with x_axis argument.

plotTimeSeries(us_covid_data)
# defaults: plot cases on the y_axis (plot_what = "cases")
#         : do not split the data by region of province, accumulate totals on day (group = "all")
#         : plot date on the x_axis (x_axis = "date")

plotTimeSeries(us_italy_covid_data, group = "region")
plotTimeSeries(us_italy_covid_data, group = "region", x_axis = "day_of_disease")
# plotTimeSeries(sars_deaths_june, plot_what = "deaths")
# plotTimeSeries(china_covid_cases_deaths, plot_what = "log_cases")
plotTimeSeries(us_italy_covid_data, plot_what = "cases_per_pop", group = "region", x_axis = "day_of_disease")
plotTimeSeries(us_italy_covid_data, plot_what = "deaths_per_cases", group = "region", x_axis = "day_of_disease")
plotTimeSeries(china_covid_cases_deaths, plot_what = "new_cases")
# plotTimeSeries(covid_data, plot_what = "growth_factor", x_axis = "day_of_disease")
# plotTimeSeries(sars_data, plot_what = "growth_factor", x_axis = "day_of_disease")
plotTimeSeries(zika_data_may_include_suspected, plot_what = "cases", x_axis = "date")