.gitignore

*.png
*.pdf
.DS_Store
.Rhistory

covid-canada.R

library(tidyverse)
library(gghighlight)
library(ggrepel)
library(scales)

fix_name_comma <- function(x) {
  name_split <- strsplit(x, ",\\s?")
  vapply(name_split, function(x) {
    if (length(x) == 1L) return(x)
    paste(x[2], x[1])
  }, FUN.VALUE = character(1))
}

data <- read_csv("https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_global.csv")

data_long <- pivot_longer(data, `1/22/20`:last_col(), names_to = "date", values_to = "cases") %>%
  filter(!is.na(cases),
         (!grepl("[Pp]rincess|[Rr]ecovered", `Province/State`) | is.na(`Province/State`)),
         `Country/Region` != "Cruise Ship", `Country/Region` != "Diamond Princess") %>%
  mutate(date = as.Date(date, format = "%m/%d/%y"),
         `Country/Region` = fix_name_comma(`Country/Region`))

max_date <- max(data_long$date)

cap <- glue::glue("Data from https://github.com/CSSEGISandData/COVID-19, {max_date}")

data_by_country <- group_by(data_long, country = `Country/Region`, date) %>%
  summarise(cases = sum(cases, na.rm = TRUE))

highlight_countries <- c("Canada", "US", "Italy", "United Kingdom", "Spain", "Australia")

(plot_countries_by_date <- filter(data_by_country, country != "China") %>%
    ggplot(aes(x = date, y = cases, colour = country)) +
    scale_x_date(date_breaks = "1 week", date_labels = "%b %d") +
    scale_y_log10(expand = c(0,0.5),
                  labels = label_comma(accuracy = 1)) +
    geom_line(size = 1.5) +
    gghighlight(country %in% highlight_countries,
                use_direct_label = FALSE,
                unhighlighted_params = list(size = 0.5)) +
    geom_text_repel(aes(label = paste0(country, " (", cases, ")")),
                    data = filter(data_by_country,
                                  country %in% highlight_countries,
                                  date == max(date)),
                    nudge_x = 1, direction = "y",
                    segment.size = 0,
                    xlim = c(max(data_by_country$date),
                             max(data_by_country$date) + 21)) +
    scale_color_brewer(palette = 2, type = "qual", guide = "none") +
    coord_cartesian(clip = "off") +
    theme_minimal() +
    theme(plot.margin = unit(c(0.5,4,0.5,0.5), "cm")) +
    labs(x = "Date",
         y = "Total number of cases (log scale)",
         title = "COVID-19 Confirmed Cases",
         caption = cap))

n_cases <- 100

data_by_country_n_days <- data_by_country %>%
  filter(cases >= n_cases) %>%
  group_by(country) %>%
  arrange(country, date) %>%
  mutate(days_since_nth_case = as.numeric(date - min(date, na.rm = TRUE))) %>%
  ungroup()


(plot_100_cases <- filter(data_by_country_n_days, country != "China") %>%
    ggplot(aes(x = days_since_nth_case, y = cases, colour = country)) +
    scale_y_log10(labels = label_comma(accuracy = 1)) +
    geom_line(size = 1.5) +
    gghighlight(country %in% c(highlight_countries, "South Korea", "Singapore", "Japan", "Iran"),
                use_direct_label = FALSE,
                unhighlighted_params = list(size = 0.5)) +
    geom_text_repel(aes(label = paste0(country, " (", cases, ")")),
                    data = filter(data_by_country_n_days,
                                  country %in% c(highlight_countries, "South Korea", "Singapore", "Japan", "Iran"),
                                  date == max(date)),
                    nudge_x = 2, direction = "both",
                    segment.size = 0) +
    coord_cartesian(clip = "off") +
    scale_color_brewer(palette = 3, type = "qual", guide = "none") +
    theme_minimal() +
    theme(plot.margin = unit(c(0.5,0.5,0.5,0.5), "cm")) +
    labs(x = glue::glue("Days since reaching {n_cases} confirmed cases"),
         y = "Total number of cases (log scale)",
         title = glue::glue("COVID-19 confirmed cases since reaching {n_cases} confirmed cases"),
         caption = cap))

data_by_country_1st_case <- data_by_country %>%
  filter(cases >= 1) %>%
  group_by(country) %>%
  arrange(country, date) %>%
  mutate(days_since_nth_case = as.numeric(date - min(date, na.rm = TRUE))) %>%
  ungroup()


(plot_1st_cases <- filter(data_by_country_1st_case, country != "China") %>%
    ggplot(aes(x = days_since_nth_case, y = cases, colour = country)) +
    scale_y_log10(labels = label_comma(accuracy = 1)) +
    geom_line(size = 1.5) +
    gghighlight(country %in% c(highlight_countries, "South Korea", "Singapore", "Japan", "Iran"),
                use_direct_label = FALSE,
                unhighlighted_params = list(size = 0.5)) +
    geom_text_repel(aes(label = paste0(country, " (", cases, ")")),
                    data = filter(data_by_country_1st_case,
                                  country %in% c(highlight_countries, "South Korea", "Singapore", "Japan", "Iran"),
                                  date == max(date)) ,
                    nudge_x = 6, direction = "both",
                    segment.size = 0,
                    xlim = c(min(data_by_country_1st_case$days_since_nth_case),
                             max(data_by_country_1st_case$days_since_nth_case) + 20)) +
    scale_color_brewer(palette = 3, type = "qual", guide = "none") +
    coord_cartesian(clip = "off") +
    theme_minimal() +
    theme(plot.margin = unit(c(0.5,4,0.5,0.5), "cm")) +
    labs(x = glue::glue("Days since first confirmed case"),
         y = "Total number of cases (log scale)",
         title = "COVID-19 confirmed cases since first confirmed case",
         caption = cap))

canada <- rename(data_long,
                 country = `Country/Region`,
                 province = `Province/State`) %>%
  filter(country == "Canada")

(plot_canada <- ggplot(canada, aes(x = date, y = cases, colour = province)) +
    geom_line(size = 1) +
    scale_x_date(date_breaks = "1 week", date_labels = "%b %d") +
    scale_y_log10(expand = c(0,1),
                  labels = label_comma(accuracy = 1)) +
    scale_color_brewer(palette = 3, type = "qual", guide = FALSE) +
    coord_cartesian(clip = "off") +
    geom_text_repel(aes(label = paste0(province, " (", cases, ")")),
                    segment.size = 0,
                    data = filter(canada, date == max(date)),
                    nudge_x = 1, direction = "y",
                    xlim = c(max(canada$date) + 1, max(canada$date) + 30)
    ) +
    theme_minimal() +
    theme(plot.margin = unit(c(0.5,6,0.5,0.5), "cm")) +
    labs(x = "Date",
         y = "Total number of cases (log scale)",
         title = "COVID-19 Confirmed Cases in Canada",
         caption = cap))


ggsave(glue::glue("countries_by_date-{max_date}.png"), plot_countries_by_date, width = 9, height = 6)
ggsave(glue::glue("100_cases-{max_date}.png"), plot_100_cases, width = 9, height = 6)
ggsave(glue::glue("1st_cases-{max_date}.png"), plot_1st_cases, width = 9, height = 6)
ggsave(glue::glue("canada-{max_date}.png"), plot_canada, width = 9, height = 6)

Thanks for sharing - I just downloaded it and it runs fine. I see you changed a couple of hardwired things - thanks. Just tried the log-axis for provinces - looking like exponential growth (though noisy).

Agreed - thanks!