R: Isarithmic Maps of Election Data (1/2)

In this tutorial I will show you I build my election maps.

Preliminaries

Before we start, we need some data and a graphic tool for the plotting:

1. I downloaded the data from  http://www.landtagswahl-bw.de.
2. I got the admin level 3 boundaries from http://www.gadm.
3.  dgd The graphics are generated with the great R-package ggplot2 developed by Hadley Wickham.

As the data has just the names and a special number of the communes, we need to get the geographic coordinates from them. For this task I used the package ggmap by which you can get the coords for your communes from google (Attention: just 2500 queries per day).

My geocode function:

getCoordsfromGoogle <- function(listofCities, printQuery=TRUE) {
    library(ggmap)
    n <- length(listofCities)
    coords <- data.frame(lon=rep(NA, n),
                         lat=rep(NA, n))
    for (i in 1:n) {
        c <- geocodeQueryCheck()
        if (printQuery) {
            print(c)
        }
        if (c > 0) {
            coords[i, ] <- geocode(listofCities[i])
        } else {
            print("Your google query is zero.")
            return(coords)
        }
    }
return(coords)
}

Now we have to prepare or election data. I omit some manually preparation of the data. First load the shape data from gadm:

load("DEU_adm3.RData")
bwk <- gadm[which(gadm$NAME_1 == "Baden-Württemberg"), ]

The election data is loaded into a data frame called lw I did a rbind with the coordinates I got by google and calculate the relative election results of the two groups:

lw$CDUrel <- (lw$CDU.CSU + lw$FDP) / lw$Gültige.Stimmen
lw$SPDrel <- (lw$SPD + lw$GRüNE + lw$DIE.LINKE) / lw$Gültige.Stimmen

lw$abs <- 100*(lw$CDUrel - lw$SPDrel)

Yet we are able to plot the first map with the raw data.

p <- ggplot() +
       geom_point(data    = lw,
                  mapping = aes(x=lon, y=lat, colour=abs),
                  size    = 3,
                  alpha   = 0.8) +
       scale_colour_gradient2(name  = "Differenz \n Wahlergebnis",
                              low   = "darkred",
                              mid   = "white",
                              high  = "blue",
                              guide = "colorbar") +
       ggtitle("Landtagswahl Baden-Württemberg 2011\n (Rohdaten)")
p + geom_path(data    = bwk,
              mapping = aes(x=long, y=lat, group=group),
              size    = 0.125)

The following graphics shows the result. Part two will follow in the next days.