In lucapresicce/DescendMethods: Descend Methods for Linear Models

knitr::opts_chunk$set(echo = TRUE)

Dataset

Use the COVID19 dataset and work on the number of deaths.

Instructions about how to get the data: https://covid19datahub.io/articles/api/r.html
Dataset description: https://covid19datahub.io/articles/doc/data.html

setwd(dirname(rstudioapi::getActiveDocumentContext()$path)) 
# install.packages("COVID19")
suppressWarnings(suppressPackageStartupMessages(library(tidyverse)))
suppressWarnings(suppressPackageStartupMessages(library(COVID19)))

x <- covid19(level = 1,  verbose = FALSE)

Request

The goal is the exercise request as well as the readability and the robustness of your code. Therefore try to elaborate a code resilient to the increase of the number of countries as well as the increase of the amount of time, for example.

Exercise 1

Create a table to present the number of total deaths by country, sorted by the cumulative number of deaths in decreasing order.

Proposed solution:

x %>% 
  filter(!is.na(iso_alpha_3)) %>% 
  select(deaths, Country = iso_alpha_3) %>% 
  group_by(Country) %>% 
  summarise(`Total deaths` = ifelse(all(is.na(deaths)),
                                  NA,
                                  sum(deaths, rm.na = T))) %>%  
  arrange(desc(`Total deaths`))

Exercises 2.

Using the COVID19 dataset, work on the number of deaths.

Aggregate data by month and use only 2020 data.

Compare the country different situations. Therefore organize the table as shown in the example below: with one row for each country (country the primary key of the table), sort the lines by the total deaths (i.e. Total Deaths) like in previous table, and show this quantity split on the time interval you chose. Therefore one more column for each period (i.e. one more column for each month) in chronological order from left to right (i.e. therefore column names will be 2020-01, 2020-02, 2020-03, ...).

Example of result:

tibble(
  country = "ITA",
  `Total Deaths` = 4,
  `2020-01` = 0, 
  `2020-02` = 1, 
  `2020-03` = 3
)

Proposed solution:

suppressWarnings(suppressPackageStartupMessages(library(lubridate)))

x %>%
  filter(!is.na(iso_alpha_3)) %>%
  select(deaths, Country = iso_alpha_3, date) %>%
  filter(year(date) == 2020 )   %>%
  mutate(date_ym = format(
    floor_date(date, unit = "month"),
    format = "%Y-%m")) %>%
  group_by(Country, date_ym) %>%
  summarise(death_per_month = ifelse(all(is.na(deaths)),
                                     NA,
                                     sum(deaths, rm.na = T))) %>% 
  ungroup() %>%
  pivot_wider(names_from = date_ym, values_from = death_per_month) %>% 
  rowwise() %>% 
  mutate(`Total deaths` = ifelse(all(is.na(c_across(where(is.numeric)))),
                                     NA,
                                     sum(c_across(where(is.numeric)), na.rm = T))) %>% 
  relocate(`Total deaths`, .after = Country)