Quick plot of prison rates versus murder rate.

prison.r

library(rvest)
library(ggvis)

wiki.prison      <- html("http://en.wikipedia.org/wiki/List_of_countries_by_incarceration_rate")
wiki.prison.data <- wiki.prison                 %>%
                        html_nodes("table")     %>%
                        html_nodes("table")     %>%
                        .[[2]]                  %>%
                        html_table(fill = TRUE)

colnames(wiki.prison.data) <- c("country", "rate")

wiki.prison.data$country[grep("United States of America", wiki.prison.data$country)] <- "United States"
wiki.prison.data$rate[grep("707. See notes below.", wiki.prison.data$rate)]          <- "707"
wiki.prison.data$rate[grep("124 or 172. See notes below.", wiki.prison.data$rate)]   <- "172"
wiki.prison.data$rate <- as.numeric(wiki.prison.data$rate)

wiki.gdp.capita <- html("http://en.wikipedia.org/wiki/List_of_countries_by_GDP_(PPP)_per_capita")
wiki.gdp.capita.data <- wiki.gdp.capita         %>%
                            html_nodes("table") %>%
                            html_nodes("table") %>%
                            .[[1]]              %>%
                            html_table()

wiki.gdp.capita.data <- wiki.gdp.capita.data[ , 2:3]
colnames(wiki.gdp.capita.data) <- c("country", "gdp.per.capita")
wiki.gdp.capita.data$gdp.per.capita <- as.numeric(gsub(",", "", wiki.gdp.capita.data$gdp.per.capita))

wiki.merge <- merge(wiki.prison.data, wiki.gdp.capita.data, by.x = "country", by.y = "country")

p <- wiki.merge                                                 %>%
        ggvis( ~ log(gdp.per.capita), ~ rate, key := ~ country) %>%
        layer_points()                                          %>%
        add_tooltip(function(df) df$country)