getElectionData.R

library(rvest)
library(dplyr)
library(tidyr)
library(readr)
library(jsonlite)
library(lubridate)

### Begin scrape ###
polls <- xml2::read_html("http://projects.fivethirtyeight.com/2016-election-forecast/national-polls/") %>%
    rvest::html_nodes(xpath = "//script[contains(., 'race.model')]") %>%
    gsub(".*race.stateData = (.*);race.pathPrefix.*", "\\1", .) %>%
    jsonlite::fromJSON(., flatten = TRUE) %>%
    magrittr::extract2("polls") %>%
    tidyr::unnest(votingAnswers) %>%
    readr::type_convert()
### End scrape   ###

### Example spread ###
polls %>%
    select(id, population, sampleSize, state, startDate, endDate, choice, pct) %>%
    tidyr::spread(choice, pct) %>%
    arrange(desc(endDate)) %>%
    slice(1:10)