A first stab at making useful heatmaps from the coronavirus data

coronavirus_heatmap.R

#devtools::install_github("RamiKrispin/coronavirus")
library(coronavirus)

library(dplyr)
library(ggplot2)
library(tidyr)
library(rnaturalearth)
library(sf)


#get a world map
worldmap <- ne_countries(returnclass = "sf")

#filter data to confirmed
coronavirus_sf <- coronavirus %>%
  st_as_sf(coords = c("Long", "Lat"), crs = st_crs(worldmap)) %>%
  mutate(date_factor = as.factor(date))

#get geospatial info about where the coronavirus is
joined_corona <- st_join(coronavirus_sf, worldmap) %>%
  rename(type = type.x)

coronavirus_wide_sf <- joined_corona %>%
  #make wide for easier calculation of total # infected
  spread(type, cases) %>%
  #not sure why values_fill isn't working...
  mutate(death = ifelse(is.na(death),0,death),
         recovered = ifelse(is.na(recovered),0,recovered),
         confirmed = ifelse(is.na(confirmed),0,confirmed)) 

coronavirus_country_summary <- coronavirus_wide_sf %>%
  #get total cases by country by date
  group_by(Country.Region, date) %>%
  summarize(cases = sum(confirmed) - sum(death) - sum(recovered) ) %>%
  ungroup() %>%
  #make sure every country is represented every date
  complete(Country.Region, date, fill = list(cases = 0)) %>%
  #arrange into a sensible order
  arrange(date, desc(cases)) %>%
  #get running sum
  group_by(Country.Region) %>%
  mutate(total_cases = cumsum(cases),
         first_case = min(date[which(cases>0)])) %>%
  ungroup() %>%
  #arrange into a sensible order
  arrange(first_case, desc(total_cases)) %>%
  mutate(Country.Region = forcats::fct_inorder(Country.Region) %>% forcats::fct_rev()) 
  

#plot by country ####
ggplot(coronavirus_country_summary,
       aes(x = date, y = Country.Region, fill = log10(total_cases))) +
  geom_tile(size = 0.2, colour="white") +
  scale_fill_viridis_c(option = "A", na.value="white", 
                       breaks = 1:5, labels = c(1:5)^10,
                       guide = guide_colorbar("Total Cases")) +
  theme_minimal() +
  xlab("") + ylab("") +
  ggtitle("Progression of Total # of Cases of Coronavirus")+ 
  theme(plot.title.position = "plot")


#let's do the same by subregion ####
coronavirus_subregion_summary <- coronavirus_wide_sf %>%
  #get total cases by country by date
  group_by(subregion, date) %>%
  summarize(cases = sum(confirmed) - sum(death) - sum(recovered) ) %>%
  ungroup() %>%
  #make sure every country is represented every date
  complete(subregion, date, fill = list(cases = 0)) %>%
  #arrange into a sensible order
  arrange(date, desc(cases)) %>%
  #get running sum
  group_by(subregion) %>%
  mutate(total_cases = cumsum(cases),
         first_case = min(date[which(cases>0)])) %>%
  ungroup() %>%
  #arrange into a sensible order
  arrange(first_case, desc(total_cases)) %>%
  mutate(subregion = forcats::fct_inorder(subregion) %>% forcats::fct_rev()) %>%
  filter(!is.na(subregion)) #figure out what is here...



ggplot(coronavirus_subregion_summary,
       aes(x = date, y = subregion, fill = log10(total_cases))) +
  geom_tile(size = 0.2, colour="white") +
  scale_fill_viridis_c(option = "A", na.value="white", 
                       breaks = 1:5, labels = c(1:5)^10,
                       guide = guide_colorbar("Total Cases")) +
  theme_minimal() +
  xlab("") + ylab("") +
  ggtitle("Total # of Cases of Coronavirus Over Time", subtitle = "subregions from naturalearth subregions\ncases = confirmed - deaths - recovered")+ 
  theme(plot.title.position = "plot")