coronavirus_us_progression.R

library(coronavirus)
library(dplyr)
library(ggplot2)
library(rnaturalearth)
library(sf)
library(ggrepel)
library(wesanderson)

coronavirus <- update_coronavirus()
usa <- ne_states(country = "United States of America", returnclass = "sf")

#make some spatial objects
usa_confirmed_sf <- coronavirus %>%
  filter(Country.Region=="US") %>%
  filter(type=="confirmed") %>%
  st_as_sf(coords = c(x = "Long", y = "Lat"),
           crs = st_crs(usa))

usa_points <- st_join(usa_confirmed_sf, usa) %>%
  group_by(name, date) %>%
  summarize(cases = sum(cases))

usa_polygons <- st_join(usa, usa_confirmed_sf) %>%
  group_by(name, date) %>%
  summarize(cases = sum(cases))

usa_states <- usa_polygons %>%
  group_by(name, date) %>%
  summarize(cases = sum(cases)) %>%
  ungroup() %>%
  group_by(name) %>%
  arrange(date) %>%
  mutate(total_cases = cumsum(cases)) %>%
  ungroup() %>%
  arrange(date, desc(cases)) %>%
  mutate(name_fct = forcats::fct_rev(forcats::fct_inorder(name)))

usa_states <- usa_states[!is.na(usa_states$cases),] #complete is weird with sf objs

#who are the top 5 for easy labels
top5 <- usa_states %>%
  group_by(name_fct, name) %>%
  summarize(total_cases = sum(cases),
            date = max(date)) %>%
  ungroup() %>%
  arrange(desc(total_cases)) %>%
  slice(1L:5L)

#heatmap? weird date gaps
ggplot(usa_states,
       aes(x = date, y = name_fct, fill = total_cases)) +
  geom_tile(color = "white") +
  scale_fill_viridis_c() +
  theme_minimal()

#line chart better idea
set.seed(2019)
ggplot(usa_states,
       aes(x = date, color = name_fct, y = total_cases)) +
  geom_line() +
  scale_color_manual(values = wes_palette("Darjeeling1", n = 37, type = "continuous")) +
  theme_minimal(base_size=10) +
  guides(color = "none") +
  geom_text_repel(data = top5,
                   mapping = aes(label = name),
                   direction    = "x",
                  force = 105) +
  ylab("Total Confirmed Cases") +
  xlab("") +
  ggtitle("US Progression of Covid-19",
          subtitle = paste0("Datat from ", format(Sys.time(), "%a %b %d %Y"), " via RamiKrispin/coronavirus for R")) +
  theme(plot.title.position = "plot")