Computes platoon advantage percentage for all hitters using Retrosheet data

platoon_dist2.R

platoon_dist2 <- function(pbpdata, season, data=FALSE){
  require(dplyr)
  require(ggplot2)
  require(Lahman)

  d_PA <- filter(pbpdata, BAT_EVENT_FL==TRUE)
  
  S_left <- summarize(group_by(filter(d_PA,
                              BAT_HAND_CD=="L"), BAT_ID),
                      Opp=sum(PIT_HAND_CD=="R"),
                      Same=sum(PIT_HAND_CD=="L"))
  S_right <- summarize(group_by(filter(d_PA,
                              BAT_HAND_CD=="R"), BAT_ID),
                       Opp=sum(PIT_HAND_CD=="L"),
                       Same=sum(PIT_HAND_CD=="R"))
  
  S <- full_join(S_left, S_right, by="BAT_ID")
  names(S)[2:5] <- c("LR", "LL", "RL", "RR")
  
  # replace NA's with 0's
  
  for (j in c(2:5)){
    N.na <- sum(is.na(S[, j]))
    S[is.na(S[, j]), j] <- rep(0, N.na)
  }
  
  # identify as right-handed, left-handed, or both sides hitter
  
  S <- mutate(S,
              L = LR + LL, R = RL + RR,
              Side=ifelse(L > 0 & R > 0, "B",
                          ifelse(L > 0, "L", "R")))
  
  S <- mutate(S,
              N_opp = LR + RL,
              N = LR + LL + RL + RR,
              P_opp = 100 * N_opp / N)
  
  SM_L <- summarize(S_left, O=sum(Opp), S=sum(Same))
  SM_R <- summarize(S_right, O=sum(Opp), S=sum(Same))
  
  SM_L <- mutate(SM_L, N=O + S, Pct = 100 * O / N)
  SM_R <- mutate(SM_R, N=O + S, Pct = 100 * O / N)
  
  TH <- theme(
    plot.title = element_text(
      colour = "blue",
      size = 18,
      hjust = 0.5,
      vjust = 0.8,
      angle = 0
    )
  )
  p <- ggplot(filter(S, N >= 100),
              aes(N, P_opp, color=Side)) +
    geom_point() + geom_smooth() +
    ggtitle(paste(season, "Platoon Advantage vs Number of PA")) +
    ylab("Pct Platoon Advantage") +
    xlab("Plate Appearances") +
    geom_hline(yintercept=SM_L$Pct) +
    geom_hline(yintercept=SM_R$Pct) +
    annotate("text", x = 650, y = SM_L$Pct + 5,
             label = "Batting Left") +
    annotate("text", x = 650, y = SM_R$Pct + 5,
             label = "Batting Right") +TH
  if(data==TRUE) S else p
}