warning_time.R

library(dplyr)
library(lubridate)
library(ggplot2)
library(ggthemes)
# European Centre for Disease Control and Prevention
EUcdc <- read.csv("https://opendata.ecdc.europa.eu/covid19/casedistribution/csv",
                  stringsAsFactors = FALSE, fileEncoding = "UTF-8-BOM")

highlight_countries <- c("New_Zealand", "United_Kingdom", "United_States_of_America",
                         "Taiwan", "Sweden", "Vietnam", "Australia")
highlight <- EUcdc %>%
  mutate(ReportDate = dmy(dateRep, tz="UTC")) %>%
  filter(ReportDate > dmy("30-1-2020", tz="UTC"),
         countriesAndTerritories %in% highlight_countries) %>% 
  arrange(countriesAndTerritories, ReportDate) %>%
  group_by(countriesAndTerritories) %>%
  mutate(total_cases = cumsum(cases),
         total_deaths = cumsum(deaths),
         days = as.integer(difftime(ReportDate, dmy("30-1-2020", tz="UTC"), units="day"))) %>% 
  ungroup() %>%
  filter(total_cases > 29) %>%
  arrange(countriesAndTerritories, ReportDate) %>%
  group_by(countriesAndTerritories) %>%
  mutate(reports=n()) %>%
  ungroup() %>%
  filter(reports > 10) %>%
  group_by(countriesAndTerritories) %>%
  summarise(days_to_20cases = days[1],
            max_cases = max(cases),
            current_cases = (cases[n()] + cases[n()-1] + cases[n()-2] + cases[n()-3] +
                               cases[n()-4] + cases[n()-5] + cases[n()-6])/7,
            current_7daysmooth_vs_max_cases = current_cases/ max_cases)
  
EUcdc %>%
  mutate(ReportDate = dmy(dateRep, tz="UTC")) %>%
  filter(ReportDate > dmy("30-1-2020", tz="UTC")) %>% 
  arrange(countriesAndTerritories, ReportDate) %>%
  group_by(countriesAndTerritories) %>%
  mutate(total_cases = cumsum(cases),
         total_deaths = cumsum(deaths),
         days = as.integer(difftime(ReportDate, dmy("30-1-2020", tz="UTC"), units="day"))) %>% 
  ungroup() %>%
  filter(total_cases > 29) %>%
  arrange(countriesAndTerritories, ReportDate) %>%
  group_by(countriesAndTerritories) %>%
  mutate(reports=n()) %>%
  ungroup() %>%
  filter(reports > 10) %>%
  group_by(countriesAndTerritories) %>%
  summarise(days_to_20cases = days[1],
            max_cases = max(cases),
            current_cases = (cases[n()] + cases[n()-1] + cases[n()-2] + cases[n()-3] +
                               cases[n()-4] + cases[n()-5] + cases[n()-6])/7,
            current_7daysmooth_vs_max_cases = current_cases/ max_cases)  %>% 
  ggplot(aes(x=days_to_20cases, y=current_7daysmooth_vs_max_cases)) + 
  geom_point(colour="#AAAAAA",alpha=0.5) + theme_minimal() + 
  geom_point(data=highlight, aes(x=days_to_20cases, y=current_7daysmooth_vs_max_cases, 
                                 colour=countriesAndTerritories)) + 
  scale_colour_colorblind() + xlab("days to 30 cases starting from Jan 30") +
  ylab("current daily cases (smoothed 7 day lag) as proportion of max daily cases")