covid-19-eda.Rmd

---
title: "COVID-19 Analytics"
date: "`r format(Sys.time(), '%d %B, %Y')`"
output:
  github_document:
    toc: false
    toc_depth: 2
    fig_width: 9
    fig_height: 6 
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE, message = FALSE)
source("render-tools.R")
```


```{r import_dependecies, include=FALSE}
options(max.print = 1e3, scipen = 999, width = 1e2)
options(stringsAsFactors = F)

suppressPackageStartupMessages({
  # data manipulations
  library(dplyr)
  library(tidyr)
  library(purrr)
  library(magrittr)
  
  # convert and formatting
  library(stringr)
  library(lubridate)
  
  # tools
  library(skimr)
  
  # graphics
  library(ggplot2)
})

source("core.R")
```


#### Table of contents {#toc}

```{r toc, echo=FALSE}
render_toc("covid-19-eda.Rmd")
```


## Load datasets

### Load COVID-19 spread data

Get list of files in datasets container:

```{r}
input_data_container <- "../data/novel-corona-virus-2019-dataset.zip"

print(
  as.character(unzip(input_data_container, list = T)$Name)
)
```

Load `covid_19_data.csv` dataset:
```{r}
covid_data <- read.csv(unz(input_data_container, "covid_19_data.csv"), 
                       na.strings = c("NA", "None", ""),
                       header = T, sep = ",")

covid_data %>% sample_n(100) %>% as_tibble
```


### Load world population data

Get datasets list:

```{r}
population_data_container <- "../data/countries.csv.zip"
print(
  as.character(unzip(population_data_container, list = T)$Name)
)
```

Load `countries.csv` dataset:

```{r}
population <- read.csv(unz(population_data_container, "countries.csv"),
                             header = T, sep = ",")

population %>% arrange(name) %>% as_tibble
```




## Preprocessing datasets

### Preprocessing COVID-19 spread data

Set `area` column, processing `province_state` columns, and format dates columns:

```{r}
# repair names
names(covid_data) <- names(covid_data) %>% 
  str_replace_all(fixed("."), "_") %>%
  str_to_lower


covid_data %<>% 
  rename(
    country = country_region,
    observation_date = observationdate
  ) %>% 
  mutate(
    ## location processing
    province_state = str_trim(province_state),
    area = as.factor(
      case_when(
        province_state == "Hubei" ~ "Hubei",
        country == "US" ~ "US",
        str_detect(country, "China") ~ "China (exclude Hubei)",
        TRUE ~ "Rest of World")),
    
    ## dates processing
    observation_date = mdy(observation_date),
    last_update = parse_date_time(str_replace_all(last_update, "T", " "), 
                                  orders = c("%Y-%m-%d %H:%M:%S", "m/d/y %H:%M"))
  )
  

covid_data %>% 
  select(area, country, province_state, observation_date, confirmed) %>% 
  arrange(desc(observation_date), desc(confirmed)) %>% 
  as_tibble
```



### Preprocessing world population data

Get unmatched countries:

```{r}
not_matched_countries <- setdiff(
  unique(covid_data$country),
  unique(population$ccse_name)
)

covid_data %>% 
  filter(country %in% not_matched_countries) %>% 
  group_by(country) %>% 
  summarise(n = sum(confirmed)) %>% 
  arrange(-n)
```

Correct top of unmached countries.

```{r}
country_map <- c("Mainland China", "UK", "Czech Republic", "Taiwan", "South Korea")
names(country_map) <- c("China", "United Kingdom", "Czechia", "Taiwan*", "Korea, South")

for (ic in 1:length(country_map)) {
  covid_data[covid_data$country == country_map[[ic]], ]$country <- names(country_map)[ic]
}
```


And updated matching:

```{r}
not_matched_countries <- setdiff(
  unique(covid_data$country),
  unique(population$ccse_name)
)

covid_data %>% 
  filter(country %in% not_matched_countries) %>% 
  group_by(country) %>% 
  summarise(n = sum(confirmed)) %>% 
  arrange(-n)
```

Much better :)


```{r set_plot_params, include=FALSE}
theme_set(theme_minimal())

lab_caption <- paste0(
  "Data source: Novel Coronavirus (COVID-19) Cases provided by Johns Hopkins University Center for Systems Science. \n",
  sprintf("Last updated: %s. ", format(max(covid_data$observation_date), '%d %B, %Y')),
  "Source code: github.com/codez0mb1e/covid-2019"
)            
```


## COVID-19 worldwide spread

***Analyze COVID-19 worldwide spread.***


### Total infected, recovered, and fatal cases

```{r}
world_spread <- covid_data %>% 
  group_by(
    country, observation_date
  ) %>% 
  summarise(
    confirmed_total = sum(confirmed),
    deaths_total = sum(deaths),
    recovered_total = sum(recovered)
  ) %>% 
  mutate(
    active_total = confirmed_total - recovered_total - deaths_total
  )
```


View spread statistics:

```{r}
world_spread %>% 
  group_by(observation_date) %>% 
  summarise_at(
    vars(ends_with("_total")), sum
  ) %>% 
  mutate_at(
    vars(ends_with("_total")), 
    list("delta" = ~ format_to_percent( calc_ratio(.) ))
  ) %>% 
  arrange(desc(observation_date)) %>% 
  select(
    observation_date, 
    starts_with("active_"), starts_with("confirmed_"), starts_with("recovered_"), starts_with("deaths_")
  )
```



### Dynamics of spread

```{r worldwide_spread_over_time}
total_population <- sum(population$population)

ggplot(
  world_spread %>% 
    group_by(observation_date) %>% 
    summarise_at(
      vars(ends_with("_total")), sum
    ) %>% 
    ungroup %>% 
    mutate(
      confirmed_total = confirmed_total - recovered_total, 
      deaths_total = -deaths_total
    ) %>% 
    gather(
      key = "case_state", value = "cases_n", -c(observation_date, active_total)
    ),
  aes(x = observation_date)) +
  
  geom_col(aes(y = cases_n, fill = case_state), alpha = .75) +
  geom_smooth(aes(y = active_total), method = "loess", color = "#0080FF", size = .2) +
  
  geom_hline(aes(yintercept = total_population*2e-3), linetype = "dashed", color = "black", alpha = .33) +
  annotate(geom = "text", label = "0.2% of world population", size = 3, color = "#666666", 
           x = min(world_spread$observation_date), y = (1.05*total_population*2e-3), 
           vjust = "bottom", hjust = "left") +
  
  scale_x_date(date_labels = "%d %b", date_breaks = "14 days") +
  scale_fill_manual(element_blank(), 
                    labels = c("confirmed_total" = "Infected cases", "recovered_total" = "Recovered cases", "deaths_total" = "Fatal cases"),
                    values = c("confirmed_total" = "grey", "recovered_total" = "gold", "deaths_total" = "black")) +
  
  labs(x = "", y = "Number of cases", 
       title = "COVID-19 Worldwide Spread (over time)", 
       subtitle = "Spread over time. \nBlue line - infected cases minus recovered and fatal.", 
       caption = lab_caption) +

  theme(
    legend.position = "top",
    plot.caption = element_text(size = 8)
  )
```

```{r}
ggplot(
  world_spread %>% 
    group_by(observation_date) %>% 
    summarise_at(
      vars(ends_with("_total")), sum
    ) %>% 
    ungroup %>% 
    mutate(
      confirmed_total = confirmed_total - recovered_total, 
      deaths_total = -deaths_total
    ) %>% 
    gather(
      key = "case_state", value = "cases_n", -c(observation_date, active_total)
    ) %>% 
    filter(observation_date > max(observation_date) - months(1)),
  aes(x = observation_date)) +
  
  geom_col(aes(y = cases_n, fill = case_state), alpha = .75) +
  geom_smooth(aes(y = active_total), method = "loess", color = "#0080FF", size = .2) +
  
  geom_hline(aes(yintercept = total_population*2e-3), linetype = "dashed", color = "black", alpha = .33) +
  annotate(geom = "text", label = "0.2% of world population", size = 3, color = "#666666", 
           x = Sys.Date() - weeks(4), y = (1.05*total_population*2e-3), 
           vjust = "bottom", hjust = "left") +
  
  scale_x_date(date_labels = "%d %b", date_breaks = "5 days") +
  scale_fill_manual(element_blank(), 
                    labels = c("confirmed_total" = "Infected cases", "recovered_total" = "Recovered cases", "deaths_total" = "Fatal cases"),
                    values = c("confirmed_total" = "grey", "recovered_total" = "gold", "deaths_total" = "black")) +
  
  labs(x = "", y = "Number of cases", 
       title = "COVID-19 Worldwide Spread", 
       subtitle = "Spread for last month. \nBlue line - infected cases minus recovered and fatal.", 
       caption = lab_caption) +

  theme(
    legend.position = "top",
    plot.caption = element_text(size = 8)
  )
```


### Disease cases structure

```{r}
ggplot(
    world_spread %>% 
      group_by(observation_date) %>% 
      summarise_at(
        vars(ends_with("_total")), sum
      ) %>% 
      ungroup %>% 
      gather(
        key = "case_state", value = "cases_n", -c(observation_date, active_total)
      )
  ) +
  
  geom_col(aes(x = observation_date, y = cases_n, fill = case_state), position = "fill", alpha = .75) +
  
  scale_x_date(date_labels = "%d %b", date_breaks = "14 days") +
  scale_fill_manual(element_blank(), 
                    labels = c("confirmed_total" = "Infected cases", "deaths_total" = "Fatal cases", "recovered_total" = "Recovered cases"),
                    values = c("confirmed_total" = "grey", "deaths_total" = "black", "recovered_total" = "gold")) +
  
  scale_y_continuous(labels = scales::percent_format(accuracy = 1)) +

  labs(x = "", y = "% from the overall number of cases", 
       title = "COVID-19 Worldwide Cases Structure", 
       subtitle = "Ratio of different types of cases over time", 
       caption = lab_caption) +
  
  theme(
    legend.position = "top",
    plot.caption = element_text(size = 8)
  )

```


```{r}
ggplot(
    world_spread %>% 
      group_by(observation_date) %>% 
      summarise_at(
        vars(ends_with("_total")), sum
      ) %>% 
      ungroup %>% 
      gather(
        key = "case_state", value = "cases_n", -c(observation_date, active_total)
      ) %>% 
      filter(observation_date > max(observation_date) - months(1))
  ) +
  
  geom_col(aes(x = observation_date, y = cases_n, fill = case_state), position = "fill", alpha = .75) +
  
  scale_x_date(date_labels = "%d %b", date_breaks = "5 days") +
  scale_fill_manual(element_blank(), 
                    labels = c("confirmed_total" = "Infected cases", "deaths_total" = "Fatal cases", "recovered_total" = "Recovered cases"),
                    values = c("confirmed_total" = "grey", "deaths_total" = "black", "recovered_total" = "gold")) +
  
  scale_y_continuous(labels = scales::percent_format(accuracy = 1)) +

  labs(x = "", y = "% from the overall number of cases", 
       title = "COVID-19 Worldwide Cases Structure", 
       subtitle = "Ratio of different types of cases for last month", 
       caption = lab_caption) +
  
  theme(
    legend.position = "top",
    plot.caption = element_text(size = 8)
  )

```


### Dynamics of daily cases

Get daily dynamics of new infected and recovered cases.


World daily spread:

```{r}
world_spread_daily <- world_spread %>% 
  group_by(observation_date) %>% 
  summarise_at(vars(ends_with("_total")), sum) %>% 
  ## calc cases per day stats
  mutate_at(
    vars(ends_with("_total")),
    list("per_day" = ~ (. - lag(.)))
  ) %>% 
  ungroup %>% 
  ## calс cases per day 
  mutate_at(
    vars(ends_with("_per_day")), 
    list(~ replace_na(., 0))
  )
  
world_spread_daily %>% 
  arrange(desc(observation_date)) %>% 
  select(observation_date, ends_with("_per_day")) %>% 
  top_n(7)
```


```{r}
ggplot(
    world_spread_daily %>% 
      transmute(
        observation_date,
        
        confirmed_total_per_day,
        active_total_per_day,
        deaths_total_per_day = -deaths_total_per_day,
        recovered_total_per_day = -recovered_total_per_day
      ) %>% 
      gather(
        key = "case_state", value = "cases_n", -c(observation_date, active_total_per_day)
      ), 
    aes(x = observation_date)
  ) +
  
  geom_col(aes(x = observation_date, y = cases_n, fill = case_state), alpha = .75) +
  geom_smooth(aes(y = active_total_per_day), method = "loess", color = "#0080FF", size = .2) +
  
  scale_x_date(date_labels = "%d %b", date_breaks = "14 days") +
  scale_fill_manual(element_blank(), 
                    labels = c("confirmed_total_per_day" = "New infected cases", "recovered_total_per_day" = "New recovered cases", "deaths_total_per_day" = "New fatal cases"),
                    values = c("confirmed_total_per_day" = "grey", "recovered_total_per_day" = "gold", "deaths_total_per_day" = "black")) +
  
  
  labs(title = "COVID-19 Worldwide Daily Spread", 
       subtitle = "Cases dynamics over time. \nBlue line - infected cases minus recovered and fatal.\nNegative values indicate recovered and fatal. cases", 
       x = "", y = "Number of new cases per day",
       caption = lab_caption) +
  
  theme(
    legend.position = "top",
    plot.caption = element_text(size = 8)
  )
```

```{r}
ggplot(
    world_spread_daily %>% 
      transmute(
        observation_date,
        
        confirmed_total_per_day,
        active_total_per_day,
        deaths_total_per_day = -deaths_total_per_day,
        recovered_total_per_day = -recovered_total_per_day
      ) %>% 
      gather(
        key = "case_state", value = "cases_n", -c(observation_date, active_total_per_day)
      ) %>% 
      filter(observation_date > max(observation_date) - months(1)), 
    aes(x = observation_date)
  ) +
  
  geom_col(aes(x = observation_date, y = cases_n, fill = case_state), alpha = .75) +
  geom_smooth(aes(y = active_total_per_day), method = "loess", color = "#0080FF", size = .2) +
  
  scale_x_date(date_labels = "%d %b", date_breaks = "5 days") +
  scale_fill_manual(element_blank(), 
                    labels = c("confirmed_total_per_day" = "New infected cases", "recovered_total_per_day" = "New recovered cases", "deaths_total_per_day" = "New fatal cases"),
                    values = c("confirmed_total_per_day" = "grey", "recovered_total_per_day" = "gold", "deaths_total_per_day" = "black")) +
  
  
  labs(title = "COVID-19 Worldwide Daily Spread", 
       subtitle = "Cases dynamics for last month. \nBlue line - infected cases minus recovered and fatal.\nNegative values indicate recovered and fatal cases.", 
       x = "", y = "Number of new cases per day",
       caption = lab_caption) +
  
  theme(
    legend.position = "top",
    plot.caption = element_text(size = 8)
  )
```



## COVID-19 spread by countries

***Analyze COVID-19 spread y countries.***


### Infected, recovered, fatal, and active cases

Calculate number of infected, recovered, fatal, and active (infected cases minus recovered and fatal) cases grouped by country:

```{r}
countries_spread <- covid_data %>% 
  ## calc number of infected, recovered, and fatal cases by country
  group_by(country, observation_date) %>% 
  summarise(
    confirmed_total = sum(confirmed),
    recovered_total = sum(recovered),
    deaths_total = sum(deaths)
  ) %>% 
  mutate(
    active_total = confirmed_total - (recovered_total + deaths_total)
  ) %>% 
  ungroup() %>% 
  
  ## calc number of days since 100 infected case
  inner_join(
    covid_data %>% 
      group_by(country) %>% 
      filter(confirmed > 100) %>% 
      summarise(since_100_confirmed_date = min(observation_date)),
    by = "country"
  ) %>%
  mutate(
    n_days_since_100_confirmed = observation_date %>% difftime(since_100_confirmed_date, units = "days") %>% as.numeric
  ) %>% 
  filter(n_days_since_100_confirmed >= 0) %>% 
  
  ## calc number of days since 10th fatal case
  left_join(
    covid_data %>% 
      group_by(country, observation_date) %>% 
      summarise(deaths_total = sum(deaths)) %>% 
      filter(deaths_total > 10) %>% 
      summarise(since_10_deaths_date = min(observation_date)),
    by = "country"
  ) %>% 
  mutate(
    n_days_since_10th_deaths = observation_date %>% difftime(since_10_deaths_date, units = "days") %>% as.numeric,
    n_days_since_10th_deaths = if_else(n_days_since_10th_deaths >= 0, n_days_since_10th_deaths, NA_real_)
  ) 

```


Get countries ordered by total active cases:

```{r}
countries_spread %>% 
  mutate_at(
    vars(ends_with("_total")), 
    list("delta" = ~ format_to_percent( calc_ratio(.) ))
  ) %>% 
  arrange(
    desc(observation_date), desc(active_total)
  ) %>% 
  select(
    country, observation_date, 
    starts_with("active_"), starts_with("confirmed_"), starts_with("recovered_"), starts_with("deaths_")
  )
```



```{r}
countries_spread_top <- countries_spread %>% 
  group_by(country) %>% 
  arrange(desc(observation_date)) %>% 
  filter(row_number() == 1) %>% 
  ungroup %>% 
  top_n(5, confirmed_total) %>% 
  select(country, ends_with("_total"))

countries_spread_top$country <- factor(countries_spread_top$country,
                                       levels = countries_spread_top$country[order(countries_spread_top$confirmed_total)])


ggplot(
   countries_spread_top %>% 
    gather(
      key = "case_state", value = "cases_n", -c(country, confirmed_total)
    ),
  aes(x = country)) +
  
  geom_col(aes(y = confirmed_total), fill = "#0080FF", alpha = .25) +
  geom_col(aes(y = cases_n, fill = country), show.legend = F) +
  
  coord_flip() +
  
  facet_grid(case_state ~ .,) +
  
  labs(x = "Country", y = "Number of Cases", 
       title = "COVID-19 Spread by Countries", 
       subtitle = "Spread by countries with the highest number of infected cases. \nBlue bar - all infected cases.", 
       caption = lab_caption) +
  
  theme(
    plot.caption = element_text(size = 8)
  )
  
```



### Dynamics of spread

```{r}
ggplot(
  countries_spread %>% 
    filter(country %in% unique(countries_spread_top$country)) %>% 
    mutate(
      confirmed_total = confirmed_total - recovered_total, 
      deaths_total = -deaths_total
    ) %>% 
    select(
      country, observation_date, ends_with("_total")
    ) %>% 
    gather(
      key = "case_state", value = "cases_n", -c(country, observation_date, active_total)
    ),
  aes(x = observation_date)) +
  
  geom_col(aes(y = cases_n, fill = case_state), alpha = .75) +
  geom_smooth(aes(y = active_total), method = "loess", color = "#0080FF", size = .2) +
  
  scale_x_date(date_labels = "%d %b", date_breaks = "14 days") +
  scale_fill_manual(element_blank(), 
                    labels = c("confirmed_total" = "Infected cases", "recovered_total" = "Recovered cases", "deaths_total" = "Fatal cases"),
                    values = c("confirmed_total" = "grey", "recovered_total" = "gold", "deaths_total" = "black")) +
  
  facet_grid(country ~ ., scales = "free") +
  
  labs(x = "", y = "Number of cases", 
       title = "COVID-19 Worldwide Spread (over time)", 
       subtitle = "Spread over time. \nBlue line - infected cases minus recovered and fatal.", 
       caption = lab_caption) +

  theme(
    legend.position = "top",
    plot.caption = element_text(size = 8)
  )
```


```{r}
ggplot(
  countries_spread %>% 
    filter(country %in% unique(countries_spread_top$country)) %>% 
    mutate(
      confirmed_total = confirmed_total - recovered_total, 
      deaths_total = -deaths_total
    ) %>% 
    select(
      country, observation_date, ends_with("_total")
    ) %>% 
    gather(
      key = "case_state", value = "cases_n", -c(country, observation_date, active_total)
    ) %>% 
    filter(
      observation_date > max(observation_date) - months(1)
    ),
  aes(x = observation_date)) +
  
  geom_col(aes(y = cases_n, fill = case_state), alpha = .75) +
  geom_smooth(aes(y = active_total), method = "loess", color = "#0080FF", size = .2) +
  
  scale_x_date(date_labels = "%d %b", date_breaks = "5 days") +
  scale_fill_manual(element_blank(), 
                    labels = c("confirmed_total" = "Infected cases", "recovered_total" = "Recovered cases", "deaths_total" = "Fatal cases"),
                    values = c("confirmed_total" = "grey", "recovered_total" = "gold", "deaths_total" = "black")) +
  
  facet_grid(country ~ ., scales = "free") +
  
  labs(x = "", y = "Number of cases", 
       title = "COVID-19 Spread by Countries", 
       subtitle = "Spread for last month. \nBlue line - infected cases minus recovered and fatal.", 
       caption = lab_caption) +

  theme(
    legend.position = "top",
    plot.caption = element_text(size = 8)
  )
```


### Dynamics of daily cases

Get daily dynamics of new infected and recovered cases by countries.


World daily spread:

```{r}
countries_spread_daily <- countries_spread %>% 
  group_by(country) %>% 
  arrange(observation_date) %>% 
  ## calc cases per day stats
  mutate_at(
    vars(ends_with("_total")),
    list("per_day" = ~ (. - lag(.)))
  ) %>% 
  ## calс cases per day 
  mutate_at(
    vars(ends_with("_per_day")), 
    list(~ replace_na(., 0))
  )
  
countries_spread_daily %>% 
  select(country, observation_date, ends_with("_per_day")) %>% 
  arrange(desc(observation_date))
```


```{r}
ggplot(
    countries_spread_daily %>% 
      filter(country %in% unique(countries_spread_top$country)) %>% 
      ungroup %>% 
      transmute(
        country, observation_date,
        
        confirmed_total_per_day,
        active_total_per_day,
        deaths_total_per_day = -deaths_total_per_day,
        recovered_total_per_day = -recovered_total_per_day
      ) %>% 
      gather(
        key = "case_state", value = "cases_n", -c(country, observation_date, active_total_per_day)
      ), 
    aes(x = observation_date)
  ) +
  
  geom_col(aes(x = observation_date, y = cases_n, fill = case_state), alpha = .75) +
  geom_smooth(aes(y = active_total_per_day), method = "loess", color = "#0080FF", size = .2) +
  
  scale_x_date(date_labels = "%d %b", date_breaks = "14 days") +
  scale_fill_manual(element_blank(), 
                    labels = c("confirmed_total_per_day" = "New infected cases", "recovered_total_per_day" = "New recovered cases", "deaths_total_per_day" = "New fatal cases"),
                    values = c("confirmed_total_per_day" = "grey", "recovered_total_per_day" = "gold", "deaths_total_per_day" = "black")) +
  
  facet_grid(country ~ ., scales = "free") +
  
  labs(title = "COVID-19 Daily Spread by Countries", 
       subtitle = "Cases dynamics over time. \nBlue line - new infected cases minus recovered and fatal.\nNegative values indicate recovered and fatal cases.", 
       x = "", y = "Number of new cases per day",
       caption = lab_caption) +
  
  theme(
    legend.position = "top",
    plot.caption = element_text(size = 8)
  )
```





```{r}
ggplot(
    countries_spread_daily %>% 
      filter(country %in% unique(countries_spread_top$country)) %>% 
      ungroup %>% 
      transmute(
        country, observation_date,
        
        confirmed_total_per_day,
        active_total_per_day,
        deaths_total_per_day = -deaths_total_per_day,
        recovered_total_per_day = -recovered_total_per_day
      ) %>% 
      gather(
        key = "case_state", value = "cases_n", -c(country, observation_date, active_total_per_day)
      ) %>% 
      filter(observation_date > max(observation_date) - months(1)), 
    aes(x = observation_date)
  ) +
  
  geom_col(aes(x = observation_date, y = cases_n, fill = case_state), alpha = .75) +
  geom_smooth(aes(y = active_total_per_day), method = "loess", color = "#0080FF", size = .2) +
  
  scale_x_date(date_labels = "%d %b", date_breaks = "5 days") +
  scale_fill_manual(element_blank(), 
                    labels = c("confirmed_total_per_day" = "New infected cases", "recovered_total_per_day" = "New recovered cases", "deaths_total_per_day" = "New fatal cases"),
                    values = c("confirmed_total_per_day" = "grey", "recovered_total_per_day" = "gold", "deaths_total_per_day" = "black")) +
  
  facet_grid(country ~ ., scales = "free") +
  
  labs(title = "COVID-19 Daily Spread by Countries", 
       subtitle = "Cases dynamics for last month. \nBlue line - new infected cases minus recovered and fatal.\nNegative values indicate recovered and fatal cases.", 
       x = "", y = "Number of new cases per day",
       caption = lab_caption) +
  
  theme(
    legend.position = "top",
    plot.caption = element_text(size = 8)
  )
```



### Mortality rate

```{r}
countries_mortality <- countries_spread %>% 
  mutate(
    confirmed_deaths_rate = deaths_total/confirmed_total,
    recovered_deaths_rate = deaths_total/(recovered_total + deaths_total)
  ) %>% 
  filter(n_days_since_10th_deaths >= 0)


countries_mortality %>% 
  filter(country == "US") %>% 
  arrange(desc(observation_date)) %>% 
  select(country, ends_with("_date"), recovered_total, contains("deaths_"))
```

```{r}
ggplot(
  countries_mortality %>% filter(country %in% unique(countries_spread_top$country)),
  aes(x = n_days_since_10th_deaths)
  ) +
  
  geom_area(aes(y = recovered_deaths_rate), alpha = .75, fill = "gold") +
  geom_area(aes(y = confirmed_deaths_rate), alpha = .75, fill = "black") +
  
  scale_y_continuous(labels = scales::percent_format(accuracy = .1)) +
  
  facet_grid(country ~ ., scales = "free") +
  
  labs(x = "Number of days since 10th fatal case", y = "Mortality rate (attention: different scales)", 
       title = "COVID-19 Mortality Rate", 
       subtitle = "Gold area - fatal to recovered cases ratio, grey area - fatal to infected cases ratio.", 
       caption = lab_caption) +
  
  theme(
    legend.position = "top",
    plot.caption = element_text(size = 8)
  )
```

```{r}
ggplot(
  countries_mortality %>% filter(country %in% unique(countries_spread_top$country)), 
  aes(x = n_days_since_10th_deaths)
  ) +
  
  geom_line(aes(y = confirmed_deaths_rate, color = country), size = 1, alpha = .75) +
  
  geom_hline(aes(yintercept = mean(countries_mortality$confirmed_deaths_rate)), linetype = "dashed", color = "black", alpha = .33) +
  annotate(geom = "text", label = "Mean mortality rate \n(over time, all countries)", x = 8, y = 1.1*mean(countries_mortality$confirmed_deaths_rate), vjust = 0, size = 3) +
  
  scale_y_continuous(labels = scales::percent_format(accuracy = 1)) +
  
  labs(x = "Number of days since 10th fatal case", y = "Mortality rate", 
       title = "COVID-19 Mortality Rate", 
       subtitle = "Fatal to infected cases ratio", 
       caption = lab_caption) +
  
  theme(
    legend.position = "top",
    legend.title = element_blank(),
    plot.caption = element_text(size = 8)
  )

```

```{r}
ggplot(
  countries_mortality %>% filter(country %in% unique(countries_spread_top$country)), 
  aes(x = n_days_since_10th_deaths)
  ) +
  
  geom_line(aes(y = recovered_deaths_rate, color = country), size = 1, alpha = .75) +
  
  geom_hline(aes(yintercept = mean(countries_mortality$recovered_deaths_rate, na.rm = T)), linetype = "dashed", color = "black", alpha = .33) +
  annotate(geom = "text", label = "Mean mortality rate \n(over time, all countries)", x = 8, y = 1.1*mean(countries_mortality$recovered_deaths_rate, na.rm = T), size = 3) +

  scale_y_continuous(labels = scales::percent_format(accuracy = 1)) +
  
  labs(x = "Number of days since 10th fatal case", y = "Mortality rate", 
       title = "COVID-19 Mortality Rate", 
       subtitle = "Fatal to recovered cases ratio", 
       caption = lab_caption) +
  
  theme(
    legend.position = "top",
    legend.title = element_blank(),
    plot.caption = element_text(size = 8)
  )
```



## COVID-19 spread by countries population

```{r}
population_spread <- countries_spread %>% 
  ## join w/ population dataset
  inner_join(
    population, by = c("country" = "ccse_name") 
  ) %>% 
  ## filter only countries with population greater than 1M
  filter(population > 1e6) %>% 
  ## calculate count of cases per 1M population
  mutate_at(
    vars(ends_with("_total")),
    list("per_1M" = ~ ./population*1e6)
  )
  
population_spread %>% 
  filter(country == "Russia") %>% 
  arrange(-n_days_since_100_confirmed) %>% 
  select(country, n_days_since_100_confirmed, population, starts_with("confirmed_"))
```


### TOPs countries by infected, active, and fatal cases

Calculate countries stats whose populations were most affected by the virus:

```{r}
population_spread_last_day <- population_spread %>% 
  ## get last observation for country
  group_by(country) %>% 
  arrange(desc(observation_date)) %>% 
  filter(row_number() == 1) %>% 
  ungroup() %>% 
  ## filter countries with infected cases greatet than 1K
  filter(confirmed_total > 1000)
```



#### ...by infected cases 

```{r}
population_spread_last_day %>% 
  arrange(-confirmed_total_per_1M) %>% 
  select(
    country, population,
    confirmed_total, confirmed_total_per_1M,
    n_days_since_100_confirmed, n_days_since_10th_deaths
  )
```


#### ...by active cases 

```{r}
population_spread_last_day %>% 
  arrange(-active_total_per_1M) %>% 
  select(
    country, population,
    active_total, active_total_per_1M,
    n_days_since_100_confirmed, n_days_since_10th_deaths
  )
```


#### ...by fatal cases 

```{r}
population_spread_last_day %>% 
  arrange(-deaths_total_per_1M) %>% 
  select(
    country, population,
    deaths_total, deaths_total_per_1M,
    n_days_since_100_confirmed, n_days_since_10th_deaths
  )
```



### Active cases per 1 million population vs number of days since 100th infected case

Select countries to monitoring:

```{r}
top_n <- 10

observed_countries <- population_spread_last_day %>% 
  arrange(desc(active_total_per_1M)) %>% 
  top_n(top_n, active_total_per_1M) %>% 
  select(country) %>% 
  as_vector %>% 
  sort


# Add countries that we are interested in
observed_countries <- c(observed_countries, "Russia", "US", "United Kingdom", "Mainland China") %>% unique
observed_countries 
```


```{r active-cases-dynamics, warning=FALSE}
active_cases_since_100st_confirmed <- population_spread %>% 
  filter(country %in% observed_countries) %>% 
  left_join(
    population_spread_last_day %>% transmute(country, active_total_per_1M_last = active_total_per_1M, n_days_since_100_confirmed),
    by = c("country", "n_days_since_100_confirmed")
  ) %>% 
  mutate(
    double_every_14d = (1 + 1/14)^n_days_since_100_confirmed, # double every 2 weeks
    double_every_7d = (1 + 1/7)^n_days_since_100_confirmed, # double every week
    double_every_3d = (1 + 1/3)^n_days_since_100_confirmed, # double every 3 days
    double_every_2d = (1 + 1/2)^n_days_since_100_confirmed # double every 2 days
  )


ggplot(active_cases_since_100st_confirmed, aes(x = n_days_since_100_confirmed)) +
  
  geom_line(aes(y = double_every_7d), linetype = "dotted", color = "red", alpha = .65) +
  geom_line(aes(y = double_every_3d), linetype = "dotted", color = "red", alpha = .85) + 
  
  geom_line(aes(y = active_total_per_1M, color = country)) +
  geom_text(aes(y = active_total_per_1M_last, label = country, color = country),
            hjust = 0.5, vjust = 0, check_overlap = T, show.legend = F, fontface = "bold", size = 3.6) +
  
  annotate(geom = "text", label = "Cases double \nevery 3 days", x = 30, y = 2800, vjust = 0, size = 3) +
  annotate(geom = "text", label = "...every week", x = 60, y = 3100, vjust = 0, size = 3) +
  
  xlim(c(110, 210)) +
  ylim(c(0, max(active_cases_since_100st_confirmed$active_total_per_1M))) +
  
  labs(x = "Number of days since 100th infected case", y = "Active cases per 1M population", 
       title = "COVID-19 Аctive Сases Dynamics", 
       subtitle = "Number of active cases per 1 million population since 100th infected case", 
       caption = lab_caption) +

  theme(
    legend.position = "bottom",
    legend.title = element_blank(),
    plot.caption = element_text(size = 8)
  )


rm(active_cases_since_100st_confirmed)
```


### Active cases per 1 million population vs number of days since 10th fatal case

```{r warning=FALSE}

active_cases_since_10th_death <- population_spread %>% 
  filter(country %in% observed_countries) %>% 
  left_join(
    population_spread_last_day %>% transmute(country, active_total_per_1M_last = active_total_per_1M, n_days_since_10th_deaths),
    by = c("country", "n_days_since_10th_deaths")
  ) %>% 
  mutate(
    double_every_14d = (1 + 1/14)^n_days_since_10th_deaths, # double every 2 weeks
    double_every_7d = (1 + 1/7)^n_days_since_10th_deaths, # double every week
    double_every_3d = (1 + 1/3)^n_days_since_10th_deaths, # double every 3 days
    double_every_2d = (1 + 1/2)^n_days_since_10th_deaths # double every 2 days
  )


ggplot(
  active_cases_since_10th_death,
  aes(x = n_days_since_10th_deaths)
  ) +
  
  geom_line(aes(y = double_every_7d), linetype = "dotted", color = "red", alpha = .65) +
  geom_line(aes(y = double_every_3d), linetype = "dotted", color = "red", alpha = .75) + 
  geom_line(aes(y = double_every_2d), linetype = "dotted", color = "red", alpha = .85) + 
  
  geom_line(aes(y = active_total_per_1M, color = country)) +
  geom_text(aes(y = active_total_per_1M_last + 20, label = country, color = country), 
            hjust = 0.5, vjust = 0, check_overlap = T, show.legend = F, fontface = "bold", size = 3.6) +
  
  annotate(geom = "text", label = "Cases double \n every 2 days", x = 18, y = 1600, vjust = 0, size = 3.1) +
  annotate(geom = "text", label = "...every 3 days", x = 25, y = 1750, vjust = 0, size = 3.1) +
  annotate(geom = "text", label = "...every week", x = 40, y = 300, vjust = 0, size = 3.1) +
  
  xlim(c(110, 210)) +
  ylim(c(0, max(active_cases_since_10th_death$active_total_per_1M) + 100)) +
  
  labs(x = "Number of days since 10th fatal case", y = "Active cases per 1M population", 
       title = "COVID-19 Аctive Сases Dynamics", 
       subtitle = "Number of active cases per 1 million population since 10th fatal case", 
       caption = lab_caption) +
  
  theme(
    legend.position = "bottom",
    legend.title = element_blank(),
    plot.caption = element_text(size = 8)
  )


rm(active_cases_since_10th_death)
```
 

*Take Care and Stay Healthy!*