World Bank Data API in R#
by Vishank Patel and Adam M. Nguyen
Documentation:
See the World Bank API documentation
These recipe examples were tested on March 24, 2023.
Setup#
# Load Packages
library(tidyverse) #ggplot2
library(dplyr) #tibbles
library(purrr) #turning into character
library(httr) #GET()
library(jsonlite) #converting to JSON
# define root World Bank API
urlRoot <- "https://api.worldbank.org/v2/"
1. Get list of country iso2Codes and names#
For obtaining data from the World Bank API, it is helpful to first obtain a list of country codes and names.
countryURL <- paste0(urlRoot,"country?format=json&per_page=500") # Create URL we are querying
raw_country_data <- GET(countryURL) # Use 'GET()' to retrieve info
prelim_country_data <- # Reading Data to R
fromJSON( # Converts JSON data to R objects
rawToChar(raw_country_data$content), flatten = TRUE) # Reads raw 8 bit data to chars
# To view try 'view(prelim_country_data)'
country_data <- prelim_country_data[[2]] # Retrieve only country data frame
country_data[1:10,1:5] # Display first 5 features of first 10 countries from country_data
## id iso2Code name capitalCity longitude
## 1 ABW AW Aruba Oranjestad -70.0167
## 2 AFE ZH Africa Eastern and Southern
## 3 AFG AF Afghanistan Kabul 69.1761
## 4 AFR A9 Africa
## 5 AFW ZI Africa Western and Central
## 6 AGO AO Angola Luanda 13.242
## 7 ALB AL Albania Tirane 19.8172
## 8 AND AD Andorra Andorra la Vella 1.5218
## 9 ARB 1A Arab World
## 10 ARE AE United Arab Emirates Abu Dhabi 54.3705
Extract Country Codes#
countryIso2Code <- as.list(country_data$iso2Code) # Extract iso2Codes
length(countryIso2Code)# Display Length
## [1] 299
head(countryIso2Code,n=10) # Display first 10
## [[1]]
## [1] "AW"
##
## [[2]]
## [1] "ZH"
##
## [[3]]
## [1] "AF"
##
## [[4]]
## [1] "A9"
##
## [[5]]
## [1] "ZI"
##
## [[6]]
## [1] "AO"
##
## [[7]]
## [1] "AL"
##
## [[8]]
## [1] "AD"
##
## [[9]]
## [1] "1A"
##
## [[10]]
## [1] "AE"
Extract country names#
countryName <- as.list(country_data$name) # Extract Country Names
length(countryName)# Display Length
## [1] 299
head(countryName,n=10) # Display first 10 Country names
## [[1]]
## [1] "Aruba"
##
## [[2]]
## [1] "Africa Eastern and Southern"
##
## [[3]]
## [1] "Afghanistan"
##
## [[4]]
## [1] "Africa"
##
## [[5]]
## [1] "Africa Western and Central"
##
## [[6]]
## [1] "Angola"
##
## [[7]]
## [1] "Albania"
##
## [[8]]
## [1] "Andorra"
##
## [[9]]
## [1] "Arab World"
##
## [[10]]
## [1] "United Arab Emirates"
Store Country Codes and Names together#
countryIso2CodeName <- transpose(list(countryIso2Code,countryName))
length(countryIso2CodeName)# Display Length
## [1] 299
head(countryIso2CodeName, n=10)
## [[1]]
## [[1]][[1]]
## [1] "AW"
##
## [[1]][[2]]
## [1] "Aruba"
##
##
## [[2]]
## [[2]][[1]]
## [1] "ZH"
##
## [[2]][[2]]
## [1] "Africa Eastern and Southern"
##
##
## [[3]]
## [[3]][[1]]
## [1] "AF"
##
## [[3]][[2]]
## [1] "Afghanistan"
##
##
## [[4]]
## [[4]][[1]]
## [1] "A9"
##
## [[4]][[2]]
## [1] "Africa"
##
##
## [[5]]
## [[5]][[1]]
## [1] "ZI"
##
## [[5]][[2]]
## [1] "Africa Western and Central"
##
##
## [[6]]
## [[6]][[1]]
## [1] "AO"
##
## [[6]][[2]]
## [1] "Angola"
##
##
## [[7]]
## [[7]][[1]]
## [1] "AL"
##
## [[7]][[2]]
## [1] "Albania"
##
##
## [[8]]
## [[8]][[1]]
## [1] "AD"
##
## [[8]][[2]]
## [1] "Andorra"
##
##
## [[9]]
## [[9]][[1]]
## [1] "1A"
##
## [[9]][[2]]
## [1] "Arab World"
##
##
## [[10]]
## [[10]][[1]]
## [1] "AE"
##
## [[10]][[2]]
## [1] "United Arab Emirates"
Now we know the country iso2codes, which we can use to pull specific indicator data for countries.
2. Compile a Custom Indicator Dataset#
There are many availabe indicators: https://data.worldbank.org/indicator
We wll select three indicators for this example:
Scientific and Technical Journal Article Data = IP.JRN.ARTC.SC
Patent Applications, residents = IP.PAT.RESD
GDP per capita (current US$) Code = NY.GDP.PCAP.CD
Note that these three selected indictaors have a CC-BY 4.0 license. We will compile this indicator data for the United States (US) and United Kingdom (GB)
indicators <- list("IP.JRN.ARTC.SC", "IP.PAT.RESD", "NY.GDP.PCAP.CD")
United States (US)#
Generate the web API URLs we need for U.S.:#
# Create an Empty List
us_api_url <- c()
# Iterate through each indicator, appending to the base URL, creating a list of unique URLs
for (indicator in indicators) {
us_api_url <- append(x = us_api_url,
values = paste(urlRoot,"country/US/indicator/",indicator,"?format=json&per_page=500",sep = ""))
}
# Display URLs
us_api_url
## [1] "https://api.worldbank.org/v2/country/US/indicator/IP.JRN.ARTC.SC?format=json&per_page=500"
## [2] "https://api.worldbank.org/v2/country/US/indicator/IP.PAT.RESD?format=json&per_page=500"
## [3] "https://api.worldbank.org/v2/country/US/indicator/NY.GDP.PCAP.CD?format=json&per_page=500"
Retrieving Data#
# Create an empty list for Indicator Data to be stored
us_indicator_data <- list()
# Iterate through URLs to collect and reformat data into lists
for (url in us_api_url) {
temp_data <- tibble(fromJSON(rawToChar(GET(url)$content), flatten = TRUE)[[2]])
us_indicator_data <- append(us_indicator_data,list(temp_data)) #making a list of tibbles
}
Extracting Data#
us_journal_data <- us_indicator_data[[1]][,c("date","value")] #the first element in us_indicator_data is regarding journal publications
head(us_journal_data,n=10)
## # A tibble: 10 × 2
## date value
## <chr> <dbl>
## 1 2021 NA
## 2 2020 NA
## 3 2019 NA
## 4 2018 422808.
## 5 2017 432216.
## 6 2016 427265.
## 7 2015 429989.
## 8 2014 433192.
## 9 2013 429570.
## 10 2012 427997.
us_patent_data <- us_indicator_data[[2]][,c("date","value")] # Takes all rows but 2nd column
head(us_patent_data,n=10)
## # A tibble: 10 × 2
## date value
## <chr> <int>
## 1 2021 NA
## 2 2020 269586
## 3 2019 285113
## 4 2018 285095
## 5 2017 293904
## 6 2016 295327
## 7 2015 288335
## 8 2014 285096
## 9 2013 287831
## 10 2012 268782
us_GDP_data <- us_indicator_data[[3]][,c("date","value")]
head(us_GDP_data)
## # A tibble: 6 × 2
## date value
## <chr> <dbl>
## 1 2021 70249.
## 2 2020 63531.
## 3 2019 65120.
## 4 2018 62823.
## 5 2017 59908.
## 6 2016 57867.
# Create a dataframe using retrieved data
us_data <- as.data.frame(c(us_journal_data,us_patent_data[2],us_GDP_data[2]),
col.names= c("Year","Scientific and Technical Journal Article Data", "Patent Applications, residents","GDP per capita (current US$) Code")) # Set column names
head(us_data)
## Year Scientific.and.Technical.Journal.Article.Data
## 1 2021 NA
## 2 2020 NA
## 3 2019 NA
## 4 2018 422807.7
## 5 2017 432216.5
## 6 2016 427264.6
## Patent.Applications..residents GDP.per.capita..current.US...Code
## 1 NA 70248.63
## 2 269586 63530.63
## 3 285113 65120.39
## 4 285095 62823.31
## 5 293904 59907.75
## 6 295327 57866.74
United Kingdom (GB)#
Now we can repeat the same process to find the relevant information for the United Kingdom indicated by the country code “GB”. As you will see, much of the code is the same, so if needed, reference the United States example.
uk_api_url <- c()
for (indicator in indicators) {
uk_api_url <- append(x = uk_api_url,
values = paste(urlRoot,"country/GB/indicator/",indicator,"?format=json&per_page=500",sep = ""))
}
uk_api_url
## [1] "https://api.worldbank.org/v2/country/GB/indicator/IP.JRN.ARTC.SC?format=json&per_page=500"
## [2] "https://api.worldbank.org/v2/country/GB/indicator/IP.PAT.RESD?format=json&per_page=500"
## [3] "https://api.worldbank.org/v2/country/GB/indicator/NY.GDP.PCAP.CD?format=json&per_page=500"
Retrieving Data#
uk_indicator_data <- list()
for (url in uk_api_url) {
temp_data <- tibble(fromJSON(rawToChar(GET(url)$content), flatten = TRUE)[[2]])
uk_indicator_data <- append(uk_indicator_data,list(temp_data)) #making a list of tibbles
}
Extracting Data#
# Extract Data
uk_journal_data <- uk_indicator_data[[1]][,c("date","value")] #takes all rows but only two columns
uk_patent_data <- uk_indicator_data[[2]][,c("date","value")]
uk_GDP_data <- uk_indicator_data[[3]][,c("date","value")]
# Combine extracted data into a data frame
uk_data <- as.data.frame(c(uk_journal_data,uk_patent_data[2],uk_GDP_data[2]),col.names= c("Year","Scientific and Technical Journal Article Data", "Patent Applications, residents","GDP per capita (current US$) Code"))
head(uk_data)
## Year Scientific.and.Technical.Journal.Article.Data
## 1 2021 NA
## 2 2020 NA
## 3 2019 NA
## 4 2018 97680.90
## 5 2017 99128.72
## 6 2016 99366.17
## Patent.Applications..residents GDP.per.capita..current.US...Code
## 1 NA 46510.28
## 2 11990 40318.56
## 3 12061 42747.08
## 4 12865 43306.31
## 5 13301 40621.33
## 6 13876 41146.08
3. Plot Indicator Data#
Create line plots of US/UK Number of Scientific and Technical Journal Articles and Patents by year. Upon inspecting the dataset, there are no values before the year 2000 and after 2018. Hence, we will slice our data for visualizations accordingly.
# Plotting the Data
# Part [4:22] corresponds to years 2000-2018
journal_data <- tibble(dates= c(us_journal_data$date[4:22]),
us_journals=c(us_journal_data$value[4:22]),
uk_journals=c(uk_journal_data$value[4:22]))
ggplot(journal_data, aes(x = dates))+
geom_line(aes(y = us_journals, color = "US IP", group=1))+
geom_point(aes(y = us_journals, color = "US IP"))+
geom_line(aes(y = uk_journals, color = "UK IP"), group=1)+
geom_point(aes(y = uk_journals, color = "UK IP"))+
labs(title = "US vs UK Journal Publications",
x="Year",
y="Publications")+
theme(axis.text.x = element_text(angle = 45, hjust = 0.5, vjust = 0.5))
Similarly for the GDP data,
gdp_data <- tibble(dates= c(us_GDP_data$date[4:22]),
us_gdp=c(us_GDP_data$value[4:22]),
uk_gdp=c(uk_GDP_data$value[4:22]))
ggplot(gdp_data, aes(x = dates))+
geom_line(aes(y = us_gdp, color = "US GDP", group=1))+
geom_point(aes(y = us_gdp, color = "US GDP"))+
geom_line(aes(y = uk_gdp, color = "UK GDP"), group=1)+
geom_point(aes(y = uk_gdp, color = "UK GDP"))+
labs(title = "US vs UK GDP",
x="Year",
y="GDP")+
theme(axis.text.x = element_text(angle = 45, hjust = 0.5, vjust = 0.5))
Patents:
patent_data <- tibble(dates= as.numeric(c(us_patent_data$date[4:41])),
us_patents=as.numeric(c(us_patent_data$value[4:41])),
uk_patents=as.numeric(c(uk_patent_data$value[4:41])))
ggplot(patent_data, aes(x = dates))+
geom_line(aes(y = us_patents, color = "US Patents", group=1))+
geom_point(aes(y = us_patents, color = "US Patents"))+
geom_line(aes(y = uk_patents, color = "UK Patents"), group=1)+
geom_point(aes(y = uk_patents, color = "UK Patents"))+
labs(title = "US vs UK Patents",
x="Year",
y="Patents")+
theme(axis.text.x = element_text(angle = 90, hjust = 0.5, vjust = 0.5)+
scale_x_continuous(breaks=seq(1980, 2020, by = 5))
)
