File:Delegate share for the 2016 Democratic Party presidential primaries.svg

library(RCurl)
library(reshape)
library(htmltab)
library(ggplot2)
library(scales)

#get the table from the url
theurl <- getURL("https://en.wikipedia.org/wiki/Template:2016USDem", ssl.verifyPeer=FALSE)
table <- htmltab(theurl)

#keep only the useful columns and name them
df <- table[, c(1, 2, 9, 12)]
names(df) <- c("Date", "State", "Clinton", "Sanders")

#transform string into date
df$Date = as.Date(substr(df[, 1], 9, 18))

#set up two data frames:
#df2 has the delegates won per candidate and per date
#df3 has the cumulatative delegate count per candidate and per date
df2 <- data.frame(Date=unique(df$Date))
df3 <- df2
for (i in 3:dim(df)[2]) {
  df[, i] = as.numeric(df[, i])
  df2[[names(df)[i]]] = rep(NA, length(df2$Date))
  df3[[names(df)[i]]] = rep(NA, length(df2$Date))
  for (j in 1:length(df2$Date)) {
    df2[[names(df)[i]]][j] = sum(subset(df, Date==df2$Date[j])[[names(df)[i]]])
    df3[[names(df)[i]]][j] = sum(df2[[names(df)[i]]][1:j])
  }
}

df2$Total = rowSums(df2[,2:3], na.rm=TRUE)
df3$Total = rowSums(df3[,2:3], na.rm=TRUE)

#reshape data so that candidate becomes a variable
mdata <- melt(df2, id=c("Date", "Total"))
mdata2 <- melt(df3, id=c("Date", "Total"))
names(mdata) <- c("Date", "Total", "Candidate", "Delegates")
mdata[["Cumulative"]] <- mdata2[, 4]
mdata[["CTotal"]] <- mdata2[, 2]

results = na.omit(mdata)

#colors
colors <- c("#D4AA00", "#228b22")
labels = c("Clinton", "Sanders")

d = ggplot(results, aes(x=Date, y=Cumulative/CTotal, fill=Candidate, colour=Candidate)) + 
  geom_area() + 
  scale_color_manual(values=colors, labels=labels) + 
  scale_fill_manual(values=colors, labels=labels) + 
  scale_y_continuous(labels=percent) + 
  labs(x="Date", y="% Pledged delegates", title="Share of delegates",
       fill="Candidate", colour="Candidate")

#display plot
d