This script pulls down information on all Nobel prize winners

get_nobel_data.R

# This R script pulls down information on all Nobel prize winners
# from the official Nobel Prize API and selects a subset of this 
# information that is then saves to a CSV-file.
# By Rasmus Bååth, 2020, in the public domain.

library(tidyverse)
library(httr)

nobel_response <- GET("http://api.nobelprize.org/2.0/laureates?limit=100000&format=json")
nobel_raw <- as_tibble(content(nobel_response, simplifyVector = TRUE, flatten = TRUE)$laureates)

nobel <- nobel_raw %>% 
  select(-starts_with("links.")) %>% 
  unnest(nobelPrizes) %>% 
  rowwise() %>% 
  mutate(
    organization_name = ifelse(is.null(affiliations$nameNow.en), NA, affiliations$nameNow.en[1]),
    organization_city = ifelse(is.null(affiliations$cityNow.en), NA, affiliations$cityNow.en[1]),
    organization_country = ifelse(is.null(affiliations$countryNow.en), NA, affiliations$countryNow.en[1])
  ) %>% 
  ungroup() %>% 
  mutate(
    awardYear = as.integer(awardYear),
    id = as.integer(id),
    birth.date = as.Date(str_replace_all(birth.date, "-00", "-01")),
    death.date = as.Date(str_replace_all(death.date, "-00", "-01")),
    prize = paste(categoryFullName.en, awardYear),
    laureate_type = ifelse(!is.na(knownName.en), "Individual", "Organization"),
    full_name = coalesce(knownName.en, orgName.en),
    prize_share = recode(portion, "1" = "1/1"),
    category.en = recode(category.en, 
      "Physiology or Medicine" = "Medicin",
      "Economic Sciences" = "Economics")
  ) %>% 
  select(
    year = awardYear,
    category = category.en,
    prize,
    motivation = motivation.en,
    prize_share,
    laureate_id = id,
    laureate_type,
    full_name,
    birth_date = birth.date,
    birth_city = birth.place.cityNow.se, 
    birth_country = birth.place.countryNow.en,
    gender, 
    organization_name,
    organization_city,
    organization_country ,
    death_date = death.date,
    death_city = death.place.cityNow.en,
    death_country = death.place.countryNow.en) %>% 
  arrange(year, category)
write_csv(nobel, "scripts/nobel.csv")