#  Based on the code created by Vincent Vu
#  This work is licensed under a Creative Commons 
#  Attribution-NonCommercial-ShareAlike 3.0 Unported License.
# 

# golfdata.Rdata contains a single object, 'golfdata', that is a list with
# 3 components: 'leaderboard', 'scorecard', 'course'
load('golfdata.RData')

# ==============================================
# = Look at an individual player's performance =
# ==============================================

# Extract Tiger Woods' score card for the 4th round and calculate his running
# over/under par score
df <- subset(golfdata$scorecard, 
             player == 'Tiger Woods' & round == 4)
tiger <- as.numeric(df[, 1:18])
tiger <- cumsum(tiger - golfdata$course$par)


plot(tiger, type = 'l')

# Extract Tiger Wood's score card and reorder the rows by the round (1 to 4)
df <- subset(golfdata$scorecard, 
             player == 'Tiger Woods')
tiger <- as.matrix(df[order(df$round), 1:18])

# Convert from an 18 x 4 matrix to a vector of length 18*4, rowwise
tiger <- as.vector(t(tiger))

# Calculate Tiger's running score across the 4 rounds of the tournament
tiger <- cumsum(tiger - rep(golfdata$course$par, 4))

plot(tiger, type = 'l')

# ==========================================================================
# = Abstract the analysis so that it can be easily repeated for any player =
# ==========================================================================
runningTotal <- function(df, par = golfdata$course$par) {
    # Reorder the rows of the data frame by the round number 
    # (so that the scores are in chronological order)
    # and extract the scores as a matrix
    x <- as.matrix(df[order(df$round), 1:18])
    n <- nrow(x)

    # Convert from an 18 x n matrix to a vector of length 18*n, rowwise
    x <- as.vector(t(x))

    # Calculate the running over/under score
    x <- cumsum(x - rep(par, n))

    return(x)
}

# Check out Charl Schwartzel
charl <- runningTotal(subset(golfdata$scorecard, 
                             player == 'Charl Schwartzel'))
plot(charl, xlab = 'hole', ylab = 'score', type = 'l')

# ==================================================================
# = Look at all players that made the cut (played in all 4 rounds) =
# ==================================================================
madecut <- subset(golfdata$leaderboard, position != 'CUT')$player
df <- subset(golfdata$scorecard, player %in% madecut)
df <- droplevels(df)

# Two approaches to computing runningTotal for each player.
# The results are stored row-wise in a 2-dimensional array

# =====================
# = For-loop approach =
# =====================
scores <- matrix(nrow = nlevels(df$player), ncol = 18 * 4)
for (i in 1:nlevels(df$player)) {
    x <- subset(df, player == levels(df$player)[i])
    scores[i, ] <- runningTotal(x)
}
rownames(scores) <- levels(df$player)

# =====================================
# = Split, apply, combine with base R =
# =====================================
x <- split(df, df$player)
x <- lapply(x, runningTotal)
scores <- do.call(rbind, x)

# ====================
# = Plot the results =
# ====================
matplot(t(scores), xlab = 'hole', ylab = 'total to par', type = 'l')
# Add the median runing total
lines(1:ncol(scores), apply(scores, 2, median), lwd = 2)    

# ==================================================================
# = Look at all players that made the cut (played in all 4 rounds) =
# ==================================================================
madecut <- subset(golfdata$leaderboard, position != 'CUT')$player
df <- subset(golfdata$scorecard, player %in% madecut)
df <- droplevels(df)