When is the Next Election in Comoros?

The International Foundation for Electoral Systems (IFES) does the incredible service of maintaining a calendar of upcoming elections at electionguide.org.

The filter UI available on the site is great if you’re interested in a single country but what if you want to check on 40 countries? Thankfully there’s an straightforward way to do just that with R.

IFES maintains a (little advertised) RSS feed that we can parse into a data.frame and subset to look at countries of interest–in this case the island nation of Comoros. The code is available on Github and I walk through the steps below:

1. Find the feed:

When I first visited electionguide.org, I missed the RSS feed because it’s only advertised near the bottom of the site’s home page and I’d landed on the upcoming elections page via search.

I ultimaetly located the feed by searching “RSS site:electionguide.org”.

2. Parse the XML:

RSS feeds are XML data so we’ll use the R XML package to parse the feed into an R data type called XMLInternalDocument and extract data from the document using a simple helper function. For more details on XML and R, check out this great presentationby Gaston Sanchez.

library(dplyr) # general data manipulation
library(XML) # parse the xml feed
library(countrycode) # working with country names is tough so i like to convert to codes

 ## parse the raw xml from IFES's Election Guide RSS feed into the R data type XMLInternalDocument
feed_raw <- xmlParse("http://www.electionguide.org/feed/calendar/upcoming", encoding = "UTF-8")

 ## make a list of the nodes we want to pull out
node_list <- c("//item/title", "//item/link", "//item/description", "//item/pubDate")

 ## helper function using xpathSApply to extact the value from each node using xmlValue
nodeGet <- function(x) {
  x <- xpathSApply(feed_raw, x, xmlValue)
  x
}

 ## apply the function to the list of nodes and turn the result into a data.frame
elect_raw <- as.data.frame(lapply(node_list, nodeGet)) 

 ## set the variable names using the original list of nodes
names(elect_raw) <- gsub("//item/", "", node_list)

3. Clean and subset for countries of interest:

This leaves us with a messy data.frame that we’ll clean up using [dplyr](https://cran.rstudio.com/web/packages/dplyr/vignettes/introduction.html) and regular expressions. Once we have the tidy data.frame that we can subset by a vector of countries and write the data out as needed.

 ## clean up the data with some regex and dplyr
elect <- elect_raw %>%
  mutate(country = gsub(":.*", "", title),
         iso3c = countrycode(country, "country.name", "iso3c"),
         elect.type = gsub(".*:", "", title),
          # remove the html from the description
         description = gsub("<.*?>|&nbsp;", "", description),
          # because i always forget the symbols for the format argument http://statmethods.net/input/dates.html
         date = as.Date(pubDate, format = "%a, %d %b %Y")) %>%
  select(country, iso3c, elect.type, date, description, link)


 ## subset the list by a vector of countries if you have a specific focus: 
countries_interest <- countrycode(c("Comoros"), "country.name", "iso3c")

elect %>%
  filter(iso3c %in% countries_interest)

I use this script to keep tabs on elections in countries where my organization has investments. When an election takes place, we evaluate how our investee’s coverage compares with other media sources visualize the results results on our Election StoryMap.

Finally, for those that read all this way, Comoros’ next election (as of this post) is a presidential contest in April.