babip (pitcher)

babip_pitcher.R

library(data.table)
library(plyr)
library(dplyr)

master = fread("Master.csv")
master$fullname = with(master, paste(nameFirst, nameLast))
name_id_data = with(master, data.frame(name = fullname, PIT_ID = retroID))

season_babip <- function(year){
  filename = paste("all", year, ".csv", sep="")  
  dat <- fread(filename)
  fields <- fread("fields.csv")
  setnames(dat, fields$Header)

  inplay_hit = c(20,21,22) 
  # 20, 21, 22 = single, double, triple
  not_inplay = c(3, 23)
  # 3, 23 = strike out, HR

  setkey(dat, AB_FL)
  dat_AB = dat[J("T")]
  dat_AB$inplay_hit = with(dat_AB, ifelse(EVENT_CD %in% inplay_hit, 1, 0))
  dat_AB$inplay = with(dat_AB, ifelse(EVENT_CD %in% not_inplay, 0, 1))

  # calculate BABIP and avg
  dat_babip = dat_AB %.% group_by(PIT_ID) %.% dplyr::summarise(babip =sum(inplay_hit)/ sum(inplay), 
                                                               inplay_hit = sum(inplay_hit), 
                                                               inplay = sum(inplay),
                                                               avg = sum(H_FL!=0)/ length(H_FL),
                                                               hit = sum(H_FL!=0),
                                                               atbat = length(H_FL)
                                                               )

  # make name data
  babip_name_data = merge(dat_babip, name_id_data, by="PIT_ID")
  babip_name_data$year = year
  
  return(babip_name_data)
}

dat = data.table()
for(N in 2010:2013){  
  dat <- rbind.data.frame(dat, season_babip(N))
}

# plot 

player_names = c("Yu Darvish", "Hiroki Kuroda", "Hisashi Iwakuma")
player_names = c(player_names, c("Clayton Kershaw", "Bartlo Colon", "R.A. Dickey", "Cliff Lee"))
subset(name_id_data, name %in% player_names)
dat_pitchers = subset(dat, name %in% player_names)

library(ggplot2)
ggplot(data = dat_pitchers, aes(x=year, y=babip, col=name)) + geom_point(size=4) + 
  geom_line(size=1)　+ ggtitle("BABIP (pitcher)") + 
  theme(plot.title=element_text(face="bold", size=20))
  

ggplot(data = dat_pitchers, aes(x=year, y=avg, col=name)) + geom_point(size=4) + 
  geom_line(size=1)　+ ggtitle("AVG (pitcher)") + 
  theme(plot.title=element_text(face="bold", size=20))