CFR Example, US

lag_us <- 7

cases <-
  read.csv("../nytimes_covid-19-data/us-states.csv") %>%
  as_tibble() %>%
  mutate(date = as.Date(date)) %>%
  select(date, state, cases) %>%
  arrange(date, state) %>%
  pivot_wider(id_cols = c("date"), names_from = c("state"), values_from = c("cases")) %>%
  replace(., is.na(.), 0) %>%
  mutate(total = rowSums(select(., !contains("date")))) %>%
  select(date, total) %>%
  rename(total_cases = total) %>%
  mutate(new_cases = c(1, diff(total_cases))) %>%
  select(date, new_cases)

deaths <-
  read.csv("../nytimes_covid-19-data/us-states.csv") %>%
  as_tibble() %>%
  mutate(date = as.Date(date) - lag_us) %>%
  select(date, state, deaths) %>%
  arrange(date, state) %>%
  pivot_wider(id_cols = c("date"), names_from = c("state"), values_from = c("deaths")) %>%
  replace(., is.na(.), 0) %>%
  mutate(total = rowSums(select(., !contains("date")))) %>%
  select(date, total) %>%
  rename(total_deaths = total) %>%
  mutate(new_deaths = c(0, diff(total_deaths))) %>%
  select(date, new_deaths)

data_cfr <-
  cases %>%
  left_join(deaths, by = "date") %>%
  mutate(cfr_us = new_deaths / new_cases) %>%
  mutate(cfr_us = ifelse(is.finite(cfr_us), cfr_us, NA)) %>%
  mutate(cfr_us_trend = ts(cfr_us, frequency = 7) %>% mstl() %>% trendcycle) %>%
  mutate(cfr_us_trend = ifelse(is.na(cfr_us), NA, cfr_us_trend)) %>%
  filter(date >= as.Date("2020-04-01"))

g_cfr <-
  ggplot(data = data_cfr) +
  theme_classic() +
  theme(axis.text.x = element_text(angle = 45)) +
  theme(axis.text.x = element_text(vjust = 0.7)) +
  theme(axis.text.x = element_text(hjust = 0.8)) +
  theme(legend.title = element_blank()) +
  #theme(legend.position = "none") +

  geom_line(size = 1.0,  aes(x = as.Date(date), y = cfr_us_trend, color = "USA")) +

  scale_x_date(date_labels = "%m/%d") +
  labs(title = "Case Fatality Rate in the US", x = "Date", y = "Case Fatality Rate") +
  scale_y_continuous(labels = scales::percent)
print(g_cfr)