Introduction to the Coronavirus Dataset

Source: vignettes/intro_coronavirus_dataset.Rmd

The coronavirus dataset

The coronavirus dataset provides a snapshot of the daily confirmed, recovered, and death cases of the Coronavirus (the 2019 Novel Coronavirus COVID-19) by geographic location (i.e., country/province). Let’s load the dataset from the coronavirus package:

library(coronavirus)

data(coronavirus)

The dataset has the following fields:

  • date - The date of the summary
  • Province.State - The province or state, when applicable
  • Country.Region - The country or region name
  • Lat - Latitude point
  • Long- Longitude point
  • cases - the number of daily cases (corresponding to the case type)
  • type - the type of case (i.e., confirmed, death)

We can use the head and str functions to see the structure of the dataset:

head(coronavirus)
#>   Province.State Country.Region Lat Long       date cases      type
#> 1                   Afghanistan  33   65 2020-01-22     0 confirmed
#> 2                   Afghanistan  33   65 2020-01-23     0 confirmed
#> 3                   Afghanistan  33   65 2020-01-24     0 confirmed
#> 4                   Afghanistan  33   65 2020-01-25     0 confirmed
#> 5                   Afghanistan  33   65 2020-01-26     0 confirmed
#> 6                   Afghanistan  33   65 2020-01-27     0 confirmed

str(coronavirus)
#> 'data.frame':    64574 obs. of  7 variables:
#>  $ Province.State: chr  "" "" "" "" ...
#>  $ Country.Region: chr  "Afghanistan" "Afghanistan" "Afghanistan" "Afghanistan" ...
#>  $ Lat           : num  33 33 33 33 33 33 33 33 33 33 ...
#>  $ Long          : num  65 65 65 65 65 65 65 65 65 65 ...
#>  $ date          : Date, format: "2020-01-22" "2020-01-23" ...
#>  $ cases         : int  0 0 0 0 0 0 0 0 0 0 ...
#>  $ type          : chr  "confirmed" "confirmed" "confirmed" "confirmed" ...

Querying and analyzing the coronavirus dataset

We will use the dplyr and tidyr packages to query, transform, reshape, and keep the data tidy, the plotly package to plot the data and the DT package to view it:

library(dplyr)
library(tidyr)
library(plotly)
library(DT)

Cases summary

Let’s start with summarizing the total number of cases by type as of 2020-04-13 and then plot it:

total_cases <- coronavirus %>% 
  group_by(type) %>%
  summarise(cases = sum(cases)) %>%
  mutate(type = factor(type, levels = c("confirmed", "death"))) 

total_cases
#> # A tibble: 3 x 2
#>   type        cases
#>   <fct>       <int>
#> 1 confirmed 1917319
#> 2 death      119482
#> 3 <NA>       448655

You can use those numbers to derive the current worldwide death rate (precentage):

round(100 * total_cases$cases[2] / total_cases$cases[1], 2)
#> [1] 6.23
plot_ly(data = total_cases, 
        x = ~ type, 
        y = ~cases, 
        type = 'bar',
        text = ~ paste(type, cases, sep = ": "),
    hoverinfo = 'text') %>%
  layout(title = "Coronavirus - Cases Distribution",
         yaxis = list(title = "Number of Cases"),
         xaxis = list(title = "Case Type"),
         hovermode = "compare")

The next plot represents the daily number of new cases worldwide:

coronavirus %>% 
  group_by(date, type) %>%
  summarise(total = sum(cases)) %>%
  pivot_wider(names_from = type, values_from = total) %>%
  plot_ly(x = ~ date, y = ~ confirmed, 
          name = "Confirmed", 
          type = "scatter", 
          mode = "none", 
          stackgroup = "one",
          fillcolor = "#4C74C9") %>%
  add_trace(y = ~death, 
            name = "Death", 
            fillcolor = "#9E0003") %>%
  layout(title = "Covid19 Daily Cases Worldwide",
         legend = list(x = 0.1, y = 0.9),
         yaxis = list(title = "Number of Cases"),
         xaxis = list(title = "Source: Johns Hopkins University Center for Systems Science and Engineering"))

Top effected countries

The next table provides an overview of the ten countries with the highest confirmed cases. We will use the datatable function from the DT package to view the table:

confirmed_country <- coronavirus %>% 
  filter(type == "confirmed") %>%
  group_by(Country.Region) %>%
  summarise(total_cases = sum(cases)) %>%
  mutate(perc = total_cases / sum(total_cases)) %>%
  arrange(-total_cases)

confirmed_country %>%
  head(10) %>%
  datatable(rownames = FALSE,
            colnames = c("Country", "Cases", "Perc of Total")) %>%
  formatPercentage("perc", 2)

The next plot summarize the distribution of the virus outside of Chine:

coronavirus %>% 
  filter(type == "confirmed", 
         Country.Region != "China") %>%
  group_by(Country.Region) %>%
  summarise(total_cases = sum(cases)) %>%
  arrange(-total_cases) %>%
  mutate(country = factor(Country.Region, levels = Country.Region)) %>%
  ungroup() %>%
  plot_ly(labels = ~ country, 
          values = ~ total_cases,
          type = "pie",
          textposition = 'inside',
        textinfo = 'label+percent',
        insidetextfont = list(color = '#FFFFFF'),
        hoverinfo = 'text',
        text = ~ paste(country, "<br />",
                     "Number of confirmed cases: ", total_cases, sep = "")) %>%
  layout(title = "Coronavirus - Confirmed Cases (Excluding China)")

Death rates

Similarly, we can use the pivot_wider function from the tidyr package (in addition to the dplyr functions we used above) to get an overview of the three types of cases (confirmed, recovered, and death). We then will use it to derive the recovery and death rate by country. As for most of the countries, there is not enough information about the results of the confirmed cases, we will filter the data for countries with at least 25 confirmed cases and above:

coronavirus %>% 
  filter(Country.Region != "Others") %>%
  group_by(Country.Region, type) %>%
  summarise(total_cases = sum(cases)) %>%
  pivot_wider(names_from = type, values_from = total_cases) %>%
  arrange(- confirmed) %>%
  filter(confirmed >= 25) %>%
  mutate(death_rate = death / confirmed)  %>%
  datatable(rownames = FALSE,
            colnames = c("Country", "Confirmed","Death", "Death Rate")) %>%
   formatPercentage("death_rate", 2)

Note that it will be misleading to make any conclusion about the recovery and death rate. As there is no detail information about:

  • There is no measurement between the time a case was confirmed and recovery or death. This is not an apple to apple comparison, as the outbreak did not start at the same time in all the affected countries.
  • As age plays a critical role in the probability of survival from the virus, we cannot make a comparison between different cases without having more demographic information.

Diving into China

The following plot describes the overall distribution of the total confirmed cases in China by province:

coronavirus %>% 
  filter(Country.Region == "China",
         type == "confirmed") %>%
  group_by(Province.State, type) %>%
  summarise(total_cases = sum(cases)) %>%  
  pivot_wider(names_from = type, values_from = total_cases) %>%
  arrange(- confirmed) %>%
  plot_ly(labels = ~ Province.State, 
                  values = ~confirmed, 
                  type = 'pie',
                  textposition = 'inside',
                  textinfo = 'label+percent',
                  insidetextfont = list(color = '#FFFFFF'),
                  hoverinfo = 'text',
                  text = ~ paste(Province.State, "<br />",
                                 "Number of confirmed cases: ", confirmed, sep = "")) %>%
  layout(title = "Total China Confirmed Cases Dist. by Province")