Where do R's supportive members mainly come from?

r-supportive-members

library(pipeR) # https://github.com/renkun-ken/pipeR
library(rlist) # https://github.com/renkun-ken/rlist
library(rvest) # https://github.com/hadley/rvest
library(stringr) # https://github.com/hadley/stringr
# please ensure rvest is the latest dev version

Pipe("http://www.r-project.org/foundation/memberlist.html")$
  html()$ # use xpath to scrape the name list
  html_nodes(xpath = "//table[2]//td//text() | //table[3]//td//text()")$
  html_text(trim = TRUE)$
  str_match_all(".+\\s\\((.+)\\)")$ # select their nations
  list.rbind()[,2]$ # combine to matrix and select nation column
  str_split(", ")$ # some members have multiple nationalities
  unlist()$
  table()$
  sort(decreasing = TRUE)$
  head(10)$
  barplot(main = "Where do R's supportive members mainly come from?")

Thanks @daroczig for sharing, that's really cool and amazing!

Error in get(y, envir = parent.frame(), mode = "function") :
object 'html_node' of mode 'function' was not found

@pssguy, please update rvest package to the latest dev version at https://github.com/hadley/rvest, it has major API change from the early release. xpath() is no longer exported, and html_node() combines css selector and xpath selector together. html_text() extracts the text so there's no need for XML::xmlValue() here.

rvest is developed to the stage that httr is no longer required here since html() fully handles the downloading and parsing, so omit library(httr).

Thanks. It's amazing what you can produce in one second now
Did you try the SelectorGadget in conjunction with html_node?

Although some of the country names do not line up, I used the library(choroplethr) to construct a map
Is there a way to pipe this (it would be nice to have it as say a side effect so as to produce a map and plot from same basic code)?

What I did is just have obj <- your code down to table() and added

df <- data.frame(tab$value)
names(df) <- c("region","value")
choroplethr(df,"world")

guess it could be more elegant

Thanks @pssguy for introducing to me the fantastic map package! It has a set of state names which unfortunately do not support alias (like USA). Some code shows how will it be organized with Pipe():

library(pipeR) # https://github.com/renkun-ken/pipeR
library(rlist) # https://github.com/renkun-ken/rlist
library(rvest) # https://github.com/hadley/rvest
library(stringr) # https://github.com/hadley/stringr

ptable <- Pipe("http://www.r-project.org/foundation/memberlist.html")$
  html()$ # use xpath to scrape the name list
  html_node(xpath = "//table[2]//td//text() | //table[3]//td//text()")$
  html_text(trim = TRUE)$
  str_match_all(".+\\s\\((.+)\\)")$ # select their nations
  list.rbind()[,2]$ # combine to matrix and select nation column
  str_split(", ")$ # some members have multiple nationalities
  unlist()$
  table()

ptable$sort(decreasing = TRUE)$
  head(10)$
  barplot(main = "Where do R's supportive members mainly come from?")  

library(choroplethr)

ptable$data.frame()$
  # some work to transform the data
  setNames(c("region","value"))$
  choroplethr("world")

If you really want to maintain the mainstream pipeline (plotting the chart) and branch it with a map, the following code works:

library(pipeR) # https://github.com/renkun-ken/pipeR
library(rlist) # https://github.com/renkun-ken/rlist
library(rvest) # https://github.com/hadley/rvest
library(stringr) # https://github.com/hadley/stringr
library(choroplethr)

ptable <- Pipe("http://www.r-project.org/foundation/memberlist.html")$
  html()$ # use xpath to scrape the name list
  html_node(xpath = "//table[2]//td//text() | //table[3]//td//text()")$
  html_text(trim = TRUE)$
  str_match_all(".+\\s\\((.+)\\)")$ # select their nations
  list.rbind()[,2]$ # combine to matrix and select nation column
  str_split(", ")$ # some members have multiple nationalities
  unlist()$
  table()$
  .(~ Pipe(.)$
      data.frame()$
      setNames(c("region","value"))$
      choroplethr("world") -> map)$
  sort(decreasing = TRUE)$
  head(10)$
  barplot(main = "Where do R's supportive members mainly come from?")$
  .(~ print(map))

This requires pipeR v0.5 to allow -> for assignment or otherwise you can only use less elegant = :)

Thanks very much
Yes Ari has done a good job on choroplethr and it works in shiny and rmarkdown which I don't believe some of the attractive rMaps options do. There is also a prob with country aliases e.g UK in my example