42-covid19.Rmd

# Covid 19 {#covid19}

> An example of an R Notebook, rendered at `r format(as.POSIXct(Sys.time()), tz="Asia/Tokyo",usetz=TRUE)`  


## Introduction

The following site of Johns Hopkins University is famous:

* https://coronavirus.jhu.edu/map.html

In this article, we study a coronavirus data collected by Johns Hopkins University called "JHU Covid-19 global time series data". Since the original data requires reshaping, we use the data provided by RamiKrispin in the following site.

* https://github.com/RamiKrispin/coronavirus/tree/master/csv

See also the R package `coronavirus` at

* https://CRAN.R-project.org/package=coronavirus
* For installation: 

```{r eval=FALSE}
install.packages("coronavirus")
```

```{r}
library(tidyverse)
library(coronavirus)
library(owidR)
```

We can directly download and read the data from:

* https://github.com/RamiKrispin/coronavirus/raw/master/csv/coronavirus.csv

It is updated daily.

In this note, we use the original JHU data and transform it using `dplyr` in the form similar to the  Krispin's.


Our world in data also provides vairous data related to covid-19. We will study a little by using `owid_covid`.


## Krispin's Package

```{r eval = FALSE, cache = TRUE}
coronavirus_tv <- read_csv("https://github.com/RamiKrispin/coronavirus/raw/master/csv/coronavirus.csv")
```

```{r eval = FALSE}
write_csv(coronavirus_tv, "data/coronavirus_tv.csv")
```

```{r echo=FALSE}
coronavirus_tv <- read_csv("data/coronavirus_tv.csv")
```


```{r cache = TRUE}
COUNTRY <- "Japan"

df_tv <- select(coronavirus_tv, c(date, country, type, cases, population))

df_tv0 <- filter(df_tv, country %in% COUNTRY)
df_tv_confirmed <- filter(df_tv0, type == "confirmed")

df_tv_confirmed_pp <- mutate(df_tv_confirmed, confirmed_pp = cases*100000/population)

ggplot(df_tv_confirmed_pp) + geom_line(aes(x = date, y = confirmed_pp), col = "red") +
  geom_smooth(aes(x = date, y = confirmed_pp), formula = y~x, method="loess", span = 0.1, se=FALSE) +
  labs(x = "Date", y = "Number of Confirmed Cases per 100,000", 
       title = paste("Number of Confirmed Cases per 100,000 ", COUNTRY))
```

```{r cache = TRUE}
COUNTRIES <- c("US", "Germany", "India", "South Africa","Korea, South", "Japan")
start_date <- as.Date("2021-07-01")
end_date <- Sys.Date() 

coronavirus_tv %>% select(c(date, country, type, cases, population)) %>% 
  filter(country %in% COUNTRIES) %>% 
  filter(date >=start_date & df_tv0$date <= end_date) %>% 
  filter(type == "confirmed") %>% 
  mutate(confirmed_pp = cases*100000/population) %>% 
  ggplot() +
  geom_smooth(aes(x = date, y = confirmed_pp, color = country), formula=y~x, method="loess", se=FALSE, span=0.1) +
  labs(x = "Date", y = "Number of Confirmed Cases per 100,000", 
       title = "Number of Confirmed Cases per 100,000")
```


```{r cache = TRUE}
COUNTRIES <- c("US", "Germany", "India", "South Africa","Korea, South", "Japan")
start_date <- as.Date("2021-07-01")
end_date <- Sys.Date() 

df_tv <- select(coronavirus_tv, c(date, country, type, cases, population))

df_tv0 <- filter(df_tv, country %in% COUNTRIES)
df_tv1 <- filter(df_tv0, date >=start_date & df_tv0$date <= end_date)
df_tv1_confirmed <- filter(df_tv1, type == "confirmed")

df_tv1_confirmed_pp <- mutate(df_tv1_confirmed, confirmed_pp = cases*100000/population)

ggplot(df_tv1_confirmed_pp) +
  geom_smooth(aes(x = date, y = confirmed_pp, color = country), formula=y~x, method="loess", se=FALSE, span=0.1) +
  labs(x = "Date", y = "Number of Confirmed Cases per 100,000", 
       title = "Number of Confirmed Cases per 100,000")
```


## Data of Johns Hopkins Universiy

### Importing Raw Data

We import the original Johns Hopkins Github data.

* COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University: https://github.com/CSSEGISandData/COVID-19
* We use [time series data](https://github.com/CSSEGISandData/COVID-19/tree/master/csse_covid_19_data/csse_covid_19_time_series)

```{r cache = TRUE}
# IMPORT RAW DATA: Johns Hopkins Github data
confirmedraw <- 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")
glimpse(confirmedraw)
confirmedraw %>% slice(1:10)# %>% datatable() # Check latest date at the end of data as tibble
```

We need to tranform data. The other data are similar.

```{r cache = TRUE}
deathsraw <- read_csv("https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_deaths_global.csv")
deathsraw %>% slice(1)
```

```{r cache = TRUE}
recoveredraw <- read_csv("https://raw.githubusercontent.com/CSSEGISandData/COVID-19/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_recovered_global.csv")
recoveredraw %>% slice(1)
# Note differences in the number of rows/columns
```

### Tidying and Combining: To create country level and global combined data

#### Convert each data set from wide to long

```{r cache = TRUE}
confirmed <- confirmedraw %>% 
  dplyr::rename(province = "Province/State", country = "Country/Region", lat = "Lat", long = "Long") %>% 
  pivot_longer(-c(province, country, lat, long), names_to = "date", values_to ="confirmed") %>% 
  mutate(date = as.Date(date, "%m/%d/%y")) %>%
  group_by(province, country) %>% arrange(date) %>%
  mutate(confirmed = confirmed - lag(confirmed)) %>%
  slice(-1) %>% ungroup() %>%
  relocate(date, .before = province) %>%
  group_by(country, province) %>% 
  arrange(province, date) 
```

Check the data.

```{r cache = TRUE}
confirmed %>% filter(country == "Japan") %>%  ggplot() + geom_line(aes(x = date, y = confirmed))
```

```{r cache = TRUE}
df_tv %>% filter(country == "Japan") %>% filter(type == "confirmed") %>%  ggplot() + geom_line(aes(x = date, y = cases))
```

_The `dplyr::rename` seems to have conflict with other `rename` function._

```{r cache = TRUE}
deaths <- deathsraw %>%  
  dplyr::rename(province = "Province/State", country = "Country/Region", lat = Lat, long = Long) %>% 
  pivot_longer(-c(province, country, lat, long), names_to = "date", values_to ="death") %>% 
  mutate(date = as.Date(date, "%m/%d/%y")) %>%
  group_by(province, country) %>% arrange(date) %>%
  mutate(death = death - lag(death)) %>%
  slice(-1) %>% ungroup() %>%
  relocate(date, .before = province) %>%
  arrange(province, date) 
```

```{r cache = TRUE}
recovered <- recoveredraw %>%  
  dplyr::rename(province = "Province/State", country = "Country/Region", lat = Lat, long = Long) %>% 
  pivot_longer(-c(province, country, lat, long), names_to = "date", values_to ="recovered") %>% 
  mutate(date = as.Date(date, "%m/%d/%y")) %>%
  group_by(province, country) %>% arrange(date) %>%
  mutate(recovered = recovered - lag(recovered)) %>%
  slice(-1) %>% ungroup() %>%
  relocate(date, .before = province) %>%
  arrange(province, date)
```


#### Final data: combine all three

```{r cache = TRUE}
coronavirus_jhu <- full_join(confirmed, deaths) %>% full_join(recovered) %>%
  pivot_longer(c(confirmed, death, recovered), names_to = "cases") %>%
  arrange(cases, province, country, date)
coronavirus_jhu %>% slice(1:10) # %>% datatable()
```
### Aggregated by Countries

The list of countries classified in provinces.

```{r cache = TRUE}
coronavirus_jhu %>% filter(!is.na(province)) %>% distinct(country)
```

Check the data associated with provinces.


If we are only interested in countries, the following is a possibility.

```{r cache = TRUE}
coronavirus_jhu_country <- coronavirus_jhu %>% 
  group_by(date, country, cases) %>% 
  summarize(value = sum(value)) %>% 
  arrange(cases, country, date)
coronavirus_jhu_country # %>% datatable()
```


## `owid_covid`

owid_covid: Get the Our World in Data covid-19 dataset

```
owid_covid()
```

See the detail at the [GitHub site](https://github.com/owid/covid-19-data/tree/master/public/data#data-on-covid-19-coronavirus-by-our-world-in-data).

* Example

```{r cash=TRUE}
covid <- owid_covid()
```

```{r}
glimpse(covid)
```


```{r}
tdpm <- covid %>% drop_na(`total_deaths_per_million`) %>%
  group_by(location) %>% 
  summarize(`total_deaths_per_million` = max(`total_deaths_per_million`)) %>%
  arrange(desc(`total_deaths_per_million`))
DT::datatable(tdpm)
```

```{r}
covid %>% drop_na(`total_deaths_per_million`) %>%
  group_by(location) %>% 
  summarize(`total_deaths_per_million` = max(`total_deaths_per_million`), .groups = "drop") %>%
  arrange(desc(`total_deaths_per_million`)) %>%
  ggplot(aes(x = `total_deaths_per_million`)) + geom_histogram(bins = 30)
```

```{r}
covid %>% filter(location == "Japan") %>% arrange(desc(date))
```

```{r}
covid %>% filter(location == "Japan") %>% drop_na(new_deaths_smoothed) %>%
  ggplot(aes(date, new_deaths_smoothed)) + geom_col() + 
    labs(title=paste("Covid-19 Death/Day in Japan as of ", Sys.Date()))
```