Gap in Life expectancy between Male and Female, 2018

Daily Graph Series

Gap in Life expectancy between Male and Female

#==============================
#   Collect and clean data
#==============================

# Load some packages: 
library(tidyverse)
library(rvest)

# Extract data from link: 
link <- "https://en.wikipedia.org/wiki/List_of_countries_by_life_expectancy?fbclid=IwAR0jRtf0RacPLxVPgcuu4zgYJl9HDk01cNy6u48vvyTuoK9CC1jZH8_pZPQ"


# Extract data: 

link %>% 
  read_html() %>% 
  html_nodes(xpath = '//*[@id="mw-content-text"]/div/table[2]') %>% 
  html_table(fill = TRUE) %>% 
  .[[1]] -> data_raw


# Rename for columns: 
names(data_raw) <- c("Rank", "Country", "Both", "Female", "Male")

# Remove the first row: 
data_raw %>% filter(Rank != "Rank") -> data_raw

# Conver to numeric and remove missing data: 

data_raw %>% 
  mutate(Country = factor(Country)) %>% 
  mutate_if(is.character, as.numeric) %>% 
  mutate(Country = as.character(Country)) %>% 
  mutate(Country = case_when(str_detect(Country, "Republic of China") ~ "China", TRUE ~ Country)) %>% 
  na.omit() -> df_life


# Select some countries with biggest gap in life expectancy: 
n <- 25

df_life %>% 
  filter(!Country %in% c("Kiribati", "Eswatini")) %>% 
  mutate(lifeGap = Female - Male) %>% 
  arrange(-lifeGap) %>% 
  slice(1:n) %>% 
  arrange(Female) %>% 
  mutate(Country = factor(Country, levels = Country)) %>% 
  mutate(diffLabel = round(lifeGap, 1)) %>% 
  mutate(diffLabel = as.character(diffLabel)) %>% 
  mutate(MaleLabel = as.character(Male), FemaleLabel = as.character(Female)) %>% 
  mutate(MaleLabel = case_when(!str_detect(MaleLabel, "\\.") ~ paste0(MaleLabel, ".0"), TRUE ~ MaleLabel)) %>% 
  mutate(FemaleLabel = case_when(!str_detect(FemaleLabel, "\\.") ~ paste0(FemaleLabel, ".0"), TRUE ~ FemaleLabel)) %>% 
  mutate(diffLabel = case_when(!str_detect(diffLabel, "\\.") ~ paste0(diffLabel, ".0"), TRUE ~ diffLabel)) %>% 
  mutate(DIFF = "DIFF") -> df


# Filter Estonia: 
df %>% filter(Country == "Estonia") -> dfEstonia


#======================
#  Solution 1 for plot
#======================

library(ggalt)

h <- 0.8
h1 <- 84.2
h2 <- h1 + 1
h3 <- -1.8
maleColor <- "#2859C9"
femaleColor <- "#B92F77"
my_font <- "Ubuntu Condensed"


df %>% 
  ggplot(aes(y = Country)) + 
  geom_dumbbell(aes(x = Male, xend = Female), size = 1.5, color = "#b2b2b2", 
                colour_x = maleColor, colour_xend = femaleColor, size_x = 4.2, size_xend = 4.2) + 
  geom_text(aes(x = Male - h, label = Male), color = maleColor, family = my_font, fontface = "bold") + 
  geom_text(aes(x = Female + h, label = Female), color = femaleColor, family = my_font, fontface = "bold") + 
  geom_rect(aes(xmin = h1, xmax = h2, ymin = -Inf, ymax = Inf), fill = "#efefe3") + 
  scale_y_discrete(expand = c(0.075, 0)) + 
  geom_text(data = dfEstonia, aes(label = "DIFF", x = (h1 + h2) / 2, y = Country), 
            fontface = "bold", family = my_font, vjust = h3) +
  geom_text(data = dfEstonia, aes(label = "Male", x = Male, y = Country), size = 4, 
            fontface = "bold", family = my_font, vjust = h3, color = maleColor) + 
  geom_text(data = dfEstonia, aes(label = "Female", x = Female, y = Country), size = 4, 
            fontface = "bold", family = my_font, vjust = h3, color = femaleColor) + 
  geom_text(aes(label = diffLabel, x = (h1 + h2) / 2, y = Country), 
            fontface = "bold", family = my_font, color = "gray30") + 
  theme_minimal(base_family = my_font) + 
  theme(panel.grid.minor = element_blank()) +
  theme(plot.background = element_rect(fill = "seashell", color = NA)) + 
  theme(axis.title = element_blank()) + 
  theme(axis.text.x = element_blank()) + 
  theme(panel.grid.major.x = element_blank()) + 
  theme(plot.margin = unit(c(1.3, 1.3, 1.3, 1.3), "cm")) + 
  theme(axis.text.y = element_text(size = 13, family = my_font)) + 
  theme(plot.title = element_text(size = 22, family = my_font, color = "gray20")) + 
  theme(plot.caption = element_text(size = 11, face = "italic", color = "grey30")) + 
  theme(plot.subtitle = element_text(size = 14, color = "grey30")) + 
  labs(x = NULL, y = NULL, 
       title = "Gender gap in life expectancy across 25 countries",
       subtitle = "Note: By descending order of female life expectancy", 
       caption = "Data Source: United Nations Development Programme")