Covid-19 Project

Covid-19 Project

Investigation of the Data

In this project I used data associated with the Coronavirus pandemic from 2020. Here are three datasets named corona_confirmed.csv, corona_recovered.csv and corona_deaths.csv. For the entirety of this project, I’ll be using these datasets. You can find more recent versions of this data at Johns Hopkins’ data resource.

library(dplyr)
library(readr)
library(knitr)
library(DT)
library(plotly)

# data loading
confirmed <- read_csv("corona_confirmed.csv")
deaths <- read_csv("corona_deaths.csv")
recovered <- read_csv("corona_recovered.csv")

# Inspection of the data
head(confirmed) %>% datatable()
head(deaths) %>% datatable()
head(recovered) %>% datatable()

Looking At March 22nd, 2020

The format of these three data frames are all the same; each row contains information about the number of cases in a certain province, state, or country. Every column (other than the columns containing the latitude, longitude, and country name) represents a date. We have data starting on January 22nd, 2020 and ending on March 22nd, 2020.

Let’s find the total number of confirmed cases on March 22nd, 2020.

# the total number of cases on March 22nd
count_confirmed <- confirmed %>%
                       select(`3/22/20`) %>%
                        sum()
count_confirmed
## [1] 335955

Filtering By Values

Let’s start to filter the data a bit. Three tasks are performed :

  • How many confirmed cases are there in countries on March 22nd that are north of the equator? (If a country is north of the equator, its latitude is greater than 0)
  • How many confirmed cases are there in March 22nd in Australia?
# countries in the northern hemisphere
northern_hemisphere_confirmed <- confirmed %>%
                            filter(Lat>0.0) %>%
                            select(`3/22/20`) %>%
                            sum()
northern_hemisphere_confirmed 
## [1] 329794
# percentage
northern_hemisphere_confirmed_percent <- (northern_hemisphere_confirmed/count_confirmed)*100
northern_hemisphere_confirmed_percent
## [1] 98.16612
# Filter for Australia cases
australia_cases <- confirmed %>%
                   filter(`Country/Region` == "Australia") %>%
                   select(`3/22/20`) %>%
                   sum()
australia_cases
## [1] 1314
#percentage
australia_cases_percent <- (australia_cases / count_confirmed)*100
australia_cases_percent
## [1] 0.3911238

Grouping By Country

Here some countries have multiple rows of data. This happens when a country has information about specific states or provinces. While this information might be useful, it makes it a bit tricky to see the total number of cases by country.

New data frame is created containing one row for every Country/Region. Every column of those new rows should have the sum of the total number of cases for that country for every day.

In this case every column other than Lat, Long, and Province/State is needed. summarize_at() only works with numbers.

After creating this new data frame inspection is done. To confirm correct calculation, I found the row for Australia and confirmed the number of cases on March 22nd. That matches my results from the previous step

# Group by countries
case_by_country <- confirmed %>%
                      group_by(`Country/Region`) %>%
                      summarize_at(vars(-Lat, -Long, -`Province/State`), sum)
head(case_by_country)  %>% datatable()
# Checking calculation confirmation by Australia count
australia_count <- case_by_country %>%
                   filter(`Country/Region` == "Australia") %>%
                   select(`3/22/20`) %>%
                   sum()
australia_count
## [1] 1314

Covid-19 ## Investigating The Recovered Dataset

Same process of grouping by country using the recovered dataset is done again. - What percentage of the cases in the US have recovered on March 22nd?

# Group by countries
recovered_by_country <- recovered %>%
                      group_by(`Country/Region`) %>%
                      summarize_at(vars(-Lat, -Long, -`Province/State`), sum)
US_recovered_22 <- recovered_by_country %>%
                    filter(`Country/Region` == "US") %>%
                    select(`3/22/20`)
                    sum()
## [1] 0
US_recovered_22
## # A tibble: 1 x 1
##   `3/22/20`
##       <dbl>
## 1         0

Its a surprising result; are there really zero recovered cases in the US? Let’s take a closer look at the US row in the recovered table.

# Filtering to inspect the US row
recovered_by_country %>%
                    filter(`Country/Region` == "US")
## # A tibble: 1 x 62
##   `Country/Region` `1/22/20` `1/23/20` `1/24/20` `1/25/20` `1/26/20` `1/27/20`
##   <chr>                <dbl>     <dbl>     <dbl>     <dbl>     <dbl>     <dbl>
## 1 US                       0         0         0         0         0         0
## # … with 55 more variables: `1/28/20` <dbl>, `1/29/20` <dbl>, `1/30/20` <dbl>,
## #   `1/31/20` <dbl>, `2/1/20` <dbl>, `2/2/20` <dbl>, `2/3/20` <dbl>,
## #   `2/4/20` <dbl>, `2/5/20` <dbl>, `2/6/20` <dbl>, `2/7/20` <dbl>,
## #   `2/8/20` <dbl>, `2/9/20` <dbl>, `2/10/20` <dbl>, `2/11/20` <dbl>,
## #   `2/12/20` <dbl>, `2/13/20` <dbl>, `2/14/20` <dbl>, `2/15/20` <dbl>,
## #   `2/16/20` <dbl>, `2/17/20` <dbl>, `2/18/20` <dbl>, `2/19/20` <dbl>,
## #   `2/20/20` <dbl>, `2/21/20` <dbl>, `2/22/20` <dbl>, `2/23/20` <dbl>,
## #   `2/24/20` <dbl>, `2/25/20` <dbl>, `2/26/20` <dbl>, `2/27/20` <dbl>,
## #   `2/28/20` <dbl>, `2/29/20` <dbl>, `3/1/20` <dbl>, `3/2/20` <dbl>,
## #   `3/3/20` <dbl>, `3/4/20` <dbl>, `3/5/20` <dbl>, `3/6/20` <dbl>,
## #   `3/7/20` <dbl>, `3/8/20` <dbl>, `3/9/20` <dbl>, `3/10/20` <dbl>,
## #   `3/11/20` <dbl>, `3/12/20` <dbl>, `3/13/20` <dbl>, `3/14/20` <dbl>,
## #   `3/15/20` <dbl>, `3/16/20` <dbl>, `3/17/20` <dbl>, `3/18/20` <dbl>,
## #   `3/19/20` <dbl>, `3/20/20` <dbl>, `3/21/20` <dbl>, `3/22/20` <dbl>

It seems like the number of recovered cases is steadily increasing to 17, until March 18th, when it suddenly drops back to 0. This is surprising, and not what is expected!

When I went back to Johns Hopkins’ resource we found a note saying that the data had moved into a different file.

We could report the maximum number of confirmed and recovered cases.

# Finding the maximum number of confirmed and recovered cases
max_recovered_US <- recovered_by_country %>%
                    filter(`Country/Region` == "US") %>%
                     select(-`Country/Region`) %>%
                    max()
max_recovered_US
## [1] 17
max_case_US <- case_by_country %>%
                    filter(`Country/Region` == "US") %>%
                     select(-`Country/Region`) %>%
                    max()
max_case_US
## [1] 33272

Transposing Data Frames

By transposing we could then find the maximum value of a country by simply selecting the appropriate column and finding the maximum value in that column. Let’s try that!

# Transposing the data frame
case_by_country <- case_by_country %>% 
                       t() %>%
                        as.data.frame()
head(case_by_country) %>% datatable()

Column Name Set

library(janitor)

# Make the first row the column titles
case_by_country <- case_by_country %>%
                          row_to_names(row_number = 1 )
head(case_by_country,26)  %>% datatable()

I printed the head of the data frame thats been just created. The columns are now of type <fctr>, or factor. This was one of the side effects of rotating the data frame.I turned all of these columns back into doubles.

# Transforming the columns to numeric values
case_by_country <- case_by_country %>%
                           apply(MARGIN = 2, as.numeric) %>%
                           as.data.frame()
head(case_by_country,25)  %>% datatable()
Sars CoV-2

Sars CoV-2

Let’s once again find the maximum number of cases reported in the US.

# Find the maximum number of confirmed cases in the US
case_by_country %>%
  select(US) %>%
  max()
## [1] 33272

Visualization

I’ve build a line graph showing the number of confirmed cases over time for a particular country. To do this, first I need to add a new column to dataset to represent the date. The first day in our dataset was January 22nd. Let’s represent that as day 1. January 23rd would then be day 2, and so on.

# Adding the date column
nrow(case_by_country)
## [1] 61
case_by_country <- case_by_country %>%
                        mutate(date = 1:61)
head(case_by_country, 28) %>%datatable()
Let’s now see the number of cases in Bangladesh over the days in the dataset.

Let’s now see the number of cases in Bangladesh over the days in the dataset.

library(ggplot2)
# Creating a line graph with date on the X axis and number of cases in Bangladesh on the Y axis
BD <- case_by_country %>% ggplot(aes(x = date, y = Bangladesh)) +
                      geom_line()
ggplotly(BD)

That line of code is pretty concise. Having a column containing only the confirmed cases from a particular country made this graph relatively simple to create.

Finally, I did a bit of work to add title.

# Adding a proper title, x label, and y label
case_by_country %>% ggplot(aes(x = date, y = Bangladesh, color = Bangladesh )) +
                      geom_line() +
                      labs(x = "Number of days since January 22nd, 2020", y = "Number of confirmed cases", title = "Covid-19 Confirmed cases In Bangladesh")

  1. Author github: Emon-ProCoder7 2. Author Linked-in:Md Tabassum Hossain Emon↩