Remake of World Bank chart

life_exp.R

# Remake of http://blogs.worldbank.org/opendata/measuring-surgical-systems-new-paradigm-health-systems-strengthening

library(dplyr)
library(purrr)
library(ggplot2)
library(scales)
library(ggthemes)
library(readxl)
library(tidyr)
library(wbstats)
library(countrycode)
library(ggrepel)

get_surgery_data <- function() {
  surgery_url <- paste0("http://media.wix.com/ugd/346076_c5fd14c703784249aae", 
                        "b8d9a43df34bc.xlsx?dn=Appendix2_Ind2_WDI_2015.xlsx")
  
  tmp <- tempfile(fileext = ".xlsx")
  download.file(surgery_url, tmp, mode = "wb")
  
  surgery <- read_excel(tmp, skip = 2)
  names(surgery)[is.na(names(surgery))] <- c("country", "income_lvl", "source")
  surgery
}

surgery <- get_surgery_data() %>% 
  gather(date, value, -country, -income_lvl, -source) %>% 
  mutate(iso2c = countrycode(country, "country.name", "iso2c")) %>% 
  filter(!is.na(value)) %>% 
  arrange(country, date) %>% 
  group_by(country) %>% 
  slice(1) %>% 
  ungroup() %>% 
  select(iso2c, "surgery" = value)

life_exp <- wb("all", "SP.DYN.LE00.IN", mrv = 1) %>% 
  tbl_df() %>% 
  select(iso2c, "life_exp" = value)

pop <- wb("all", "SP.POP.TOTL", mrv = 1) %>% 
  tbl_df() %>% 
  select(iso2c, "pop" = value)

df <- reduce(list(life_exp, pop, surgery, wbcountries()), inner_join)

df$region <- gsub("\\(.*\\)", "", df$region)
df$residual <- loess(life_exp ~ surgery, df)$residuals %>% abs()

ggplot(df, aes(x = surgery, y = life_exp, size = pop, color = region)) + 
  geom_vline(xintercept = c(20, 40), color = "grey60") +
  geom_point(alpha = 0.25) +
  geom_smooth(color = "whitesmoke", alpha = 0.2) +
  geom_text_repel(aes(label = country), top_n(df, 25, residual), size = 3,
                  show.legend = FALSE) +
  scale_size_area(breaks = pretty_breaks(4), max_size = 10) +
  scale_x_log10(breaks = c(1, 20, 40, 100)) +
  scale_color_few("dark") +
  theme_light() +
  theme(text = element_text(color = "grey40"), 
        axis.title = element_text(size = rel(0.8)),
        legend.key = element_blank(),
        panel.background = element_rect(fill = "grey99"),
        plot.caption = element_text(size = rel(0.6), 
                                    margin = unit(c(5, 5, 0, 5), "mm"))) +
  guides(size = "none", color = guide_legend(override.aes = list(size = 5))) +
  labs(x = "Specialist surgical workers (per 100,000 people)",
       y = "Life expectancy at birth (years)\n",
       color = NULL,
       title = paste0("Life expectancy vs. the relative number of ", 
                      "surgical workers in a country"),
       subtitle = "Estimates around 2013",
       caption = paste0("Source: World Bank, Lancet Commission\nNote: ", 
                        "Vertical lines indicate Lancet Commission benchmarks")
  )