Question

I´m trying to produce a worldmap using ggplot2 in R. It schould be a countrybased heatmap. The data I’m working on comes from twitter and I want to show where the tweets come from. There are 2 problems:

1. the integrated data

   map_data("world")
   

   gives me a map older the 20 years (USSR).

   map_data("world2")
   

   seems damaged. or there is some ordering issue but I don’t know how to solve it.

http://schloegl.net/supersambo/world2.pdf

1. I would like to change the colour breaks and don’t know how to access it. Since Brazil is the only country easy to read it would be important to edit the breaks and show the difference between countries with less then 1000 tweets.

http://schloegl.net/supersambo/world.pdf

here is my code

    WD <- getwd()
    if (!is.null(WD)) setwd(WD)

    library(maps)
    library(plyr)
    library(ggplot2)

    twitter=read.csv("/Users/stephanschloegl/Studium/Diplomarbeit/rawData/c_userInfo/c_userInfo.csv",header=TRUE,check.names=FALSE,sep=";")

    #read geodata
    cities=read.csv("GeoWorldMap/cities.txt",header=TRUE,check.names=FALSE,sep=",")
    countries=read.csv("GeoWorldMap/countries.txt",header=TRUE,check.names=FALSE,sep=",")

    #find countries for twitter$timezone
    lista <- twitter$time_zone
    country_ids <- cities$CountryID[match(lista,cities$City)]
    country <- countries$Country[match(country_ids,countries$CountryId)]

    #FREQENCIES
    frequencies <- as.data.frame(table(country))
    names(frequencies) <- c("region","freq")
    #change 0's to NA
    frequencies$freq[frequencies$freq==0] <- NA

    #load world data
    world <- map_data("world2")
    #Delete Antarctica
    world <- subset(world,region!="Antarctica")


    #merge twitterdata and geodata
    world$tweets <- frequencies$freq[match(world$region,frequencies$region,nomatch=NA)]

    map <- qplot(long, lat, data = world, group = group,fill=tweets,geom ="polygon",ylab="",xlab="")
    #this does'nt work
    map + scale_colour_hue(name="Number of\nTweets",breaks=levels(c(10,20,100,200,1000)))
    map

Answer 1

That’s because scale_colour_hue() is for discrete scales. You have to use scale_fill_gradient() because you want to change the fill not the outline.

    map + scale_fill_gradient(name="Number of\nTweets", trans = "log",
                           breaks = c(10, 20, 100, 200, 1000))

There you go. You have also put levels in the breaks which makes no sense, the number of tweets is numeric.

You can get the new map here and follow the instructions in that post to make it work.