providencefirefighters.R

gistfile1.r

install.packages('rvest')
library(rvest)
library(dplyr)
page <- html('http://app.providencejournal.com/topics/special-reports/tables/firefigher-salaries-fy2014/firefighter-salaries.htm')
data <- page %>% 
        html_node('table') %>% 
        html_table() %>% 
        .[c(-1,-2,-3),] %>% 
        as.data.frame

data <- as.data.frame(apply(data, 2, function(x) gsub('\\$', '', x)))
data[, c('Listed Salary', 'Base pay', 'Longevity', 
         'Overtime/ Callback', 'Detail pay', 'Retro Pay', 'Other',
         'Total')] <- apply(data[, c('Listed Salary', 'Base pay', 'Longevity', 
                                      'Overtime/ Callback', 'Detail pay', 'Retro Pay', 'Other',
                                      'Total')], 2,
                             function(x) gsub(',', '', x))
data[, c('Listed Salary', 'Base pay', 'Longevity', 
         'Overtime/ Callback', 'Detail pay', 'Retro Pay', 'Other',
         'Total')] <- apply(data[, c('Listed Salary', 'Base pay', 'Longevity', 
                                      'Overtime/ Callback', 'Detail pay', 'Retro Pay', 'Other',
                                      'Total')], 2,
                             as.numeric)
library(ggplot2)
library(scales)
ggplot(data, aes(`Base pay`, `Overtime/ Callback`)) + 
  geom_point() +
  geom_vline(xintercept = 50000) +
  scale_x_continuous('Base Pay', 
                     labels = dollar) +
  scale_y_continuous('Overtime or Callback Pay', 
                     labels = dollar) +
  theme_bw() +
  ggtitle("Who Earns Overtime Pay?\nProvidence Firefighters")