Code to Visualize "Pianograms" - Winner-take-all type plots given two or more players. In this example, we plot Kasparov vs Karpov's Wins - sliced yearly, 5-year and decade-wise buckets. Code accompanies this blog post:

k_vs_k_plotting.R

  rm(list=ls())
library(ggplot2)
library(plyr)
library(reshape2)


df <- read.csv("~/RStats/data/KvsK_lifetime.csv", stringsAsFactors=F, strip.white=T)
df <- arrange(df, game) #use plyr's arrange to Sort by Game Number

#For ease drop the Opening, Event.Locale columns and  Moves Columns
df <- df[, c(1,2,5, 3)]

#Who had White? Create a new column
df$GK.white <- ifelse(df$KvsK == "Karpov vs Kasparov", 0, 1)

#Create to new column for the Winner.
df$GK.won <- 0 # we will be overwriting this column in instances where GK Won
for (g in 1:nrow(df)) {
  if(df$result[g] == "Draw") {df$GK.won[g] <- "Draw"}
  else if ((df$GK.white[g]==1 & df$result[g]==1) || (df$GK.white[g]==0 & df$result[g]==0)) {
    df$GK.won[g] <- 1
  }
}

# for the sake of plotting, we drop all the Draw Results
# df <-   df[df$GK.won != "Draw", ]
# df$GK.won <- as.numeric(df$GK.won)

#drop the game and kvsk column
df <- df[c(-1,-2)]

##Create a placeholder Data Frame
## bucket each year into Y, FiveYear and its Decade
x = seq(1970,2005, by=5)
d = seq(1970,2000, by=10)
years.df <- data.frame(
  year = 1970:2009,
  five.year = unlist(lapply(x, FUN=rep, 5)),
  decade = unlist(lapply(d, FUN=rep, 10))
)

df <- merge(years.df, df, all.x=T) #add in 2 new columns, and fill in all the year gaps.


# @ Given a data frame slice, this function determines who had more wins in that slice
moreWins <- function(df) {  
  valid.rows <- sum(!is.na(df$GK.won)) # count of 0 or 1 values
  if(valid.rows==0) {
    return("00")
  }
  gk.wins <- sum(na.omit(df$GK.won == "1")) #how many times did GK win
  gk.losses <- sum(na.omit(df$GK.won == "0")) #how many times did GK lose
  
  if(gk.wins > gk.losses) {
    return("GK")
  }
  else if(gk.wins == gk.losses) {
    return("DR")
  }
  else {
    return("AK")  
  }  
  
}




yearlywinner.df <- ddply(df, .(year), moreWins)
names(yearlywinner.df)[2] <- "winner.1"

five.year.winner.df <- ddply(df, .(five.year), moreWins)
names(five.year.winner.df)[2] <- "winner.5"

decade.winner.df <- ddply(df, .(decade), moreWins)
names(decade.winner.df)[2] <- "winner.10"


y.full  <- merge(years.df, yearlywinner.df)
fy.full <- merge(years.df, five.year.winner.df)
dec.full <- merge(years.df, decade.winner.df)

names(y.full)
plot.df <- y.full[c(1,4)]
plot.df$winner.5   <- fy.full$winner.5
plot.df$winner.10 <- dec.full$winner.10

m.df <- melt(plot.df, id.vars="year")

m.df$variable <- factor(m.df$variable,
                        levels=c("winner.1", "winner.5","winner.10" ),
                        ordered=TRUE)


## plotting - Pianograms
win_labeller <- function(var, value){
  value <- as.character(value)
  if (var=="variable") { 
    value[value=="winner.1"] <- "Yearly Winner"
    value[value=="winner.5"]   <- "5-Year Winner"
    value[value=="winner.10"]   <- "Decade Winner"
  }
  return(value)
}

head(m.df)

breaks <- seq(1970,2010,by=5)

## Use this to plot each time-period plot. (Change the fill to winner.1, winner.5 or winner.10)
p1 <- ggplot(plot.df) + geom_bar(aes(x=year, fill=winner.10), binwidth=1, color="gray90")
p1 <- p1 + labs(title="Karpov vs Kasparov, Winner by Decade")
p1 <- p1 + ylab("Year's Winner")+ scale_x_continuous(breaks=breaks) 
p1 <- p1  + theme(axis.text.y=element_blank(), legend.title=element_blank()) + scale_fill_manual(
  limits=c("00", "AK", "GK", "DR"),
  values= c("white","blue","darkgreen", "gray60"),
  labels=c("Did not play", "Anatoly Karpov", "Garry Kasparov", "Draw")
  )
p1  

p <- NULL
vlines <- seq(1980, 2010, by=10)
p <- ggplot(m.df) + geom_bar(aes(x=year, fill=value), binwidth=1, color="gray90") + facet_grid (
  variable ~ ., labeller=win_labeller
)
p <- p +   ylab("Decade, 5-year and Yearly Winners")

p <- p  + scale_fill_manual(
  limits=c("00", "AK", "GK", "DR"),
  values= c("white","blue","darkgreen", "gray70"),
  labels=c("Did not play", "Anatoly Karpov", "Garry Kasparov", "Draw")
)

p <- p  + theme(panel.background = element_blank(),
                axis.text.y=element_blank(),
                axis.ticks=element_blank()
                ) + geom_vline(xintercept = vlines, color="darkred", size=2) + 
  labs(title="Karpov vs Kasparov, Lifetime Winner-Take-All Tally") 
p

kasparov_karpov.csv