Code for Generating a Scatter Plot of Citi Bike Share Station Names

citi_bike_share.r

library(ggplot2)
library(rjson)

# Strip out enclosing object so its just an array of stations before importing into R.
# e.g. data should be of the form:
# [
#  {"id":72,"stationName":"W 52 St & 11 Av","availableDocks":14,"totalDocks":39,"latitude":40.76727216,"longitude":-73.99392888,"statusValue":"In Service","statusKey":1,"availableBikes":21,"stAddress1":"W 52 St & 11 Av","stAddress2":"","city":"","postalCode":"","location":"","altitude":"","testStation":false,"lastCommunicationTime":null,"landMark":""}, 
#  ... 
# ]
#

bike <- fromJSON(paste(readLines("bike_data.json"), collapse = ""))
bike_colnames <- names(unlist(bike)[0:17])
bike <- data.frame(matrix(unlist(bike), ncol=17, byrow=T))
colnames(bike) <- bike_colnames

qplot(data = bike, x = as.numeric(as.character(totalDocks)), y = as.numeric(as.character(availableBikes)), 
    color = as.numeric(as.character(latitude)), 
    label = stationName, geom = "text", angle = 10) + 
  scale_x_continuous("Total Docks") + 
  scale_y_continuous("# of Available Bikes") + 
  scale_color_gradient("Latitude (darker is downtown)")