# R scripts from paper
# Baseball Data at Season, Play-by-Play, and Pitch-by-Pitch   
# Levels by Jim Albert

# Section 2   Lahman Baseball Database
# Section 2.1 Description

batting=read.table("mlb_batting.dat", header=TRUE)

mg=subset(batting,
  batting$first.name=="Mark"&batting$last.name=="McGwire")

mg[,c("year","ab","h","hr")]

# Section 2.2  A Sample Analysis

hits.ab=aggregate(batting[,c("h","ab")],
  list(year=batting$year),sum,na.rm=TRUE)
avg=hits.ab$h/hits.ab$ab
plot(hits.ab$year, avg, xlab="Year", ylab="Batting Average")
lines(lowess(hits.ab$year, avg, f = 1/8))

so.ab=aggregate(batting[,c("so","ab")],
  list(year=batting$year),sum,na.rm=TRUE)
so.rate=so.ab$so/so.ab$ab
plot(so.ab$year, so.rate, xlab="Year", ylab="Strikeout Rate")
lines(lowess(so.ab$year, so.rate, f=1/8))

# Section 3   Retrosheet Data
# Section 3.1 Description

pbp=read.table("playbyplay2008.dat", header=TRUE)

roster=read.table("roster2008.dat", header=TRUE)

howard.hr=subset(pbp,pbp$date=="5100"&pbp$v_team=="PHI"&
   pbp$batter=="howar001"&pbp$inning==2)
   
howard.hr[,c("inning","outs","balls","strikes",
   "b1_runner","b2_runner","b3_runner","event")]

# Section 3.2  A Sample Analysis

with(roster, abbrev[first.name=="Derek"&last.name=="Jeter"])

jeter=subset(pbp, pbp$batter=="jeted001"&
  pbp$ab_flag==TRUE)

y=ifelse(jeter$hit_value>0,1,0)

m.avg=filter(y, rep(1/20,20))
plot(m.avg,type="l",ylab="Moving Batting Average")

# Section 4   PITCHf/x Database
# Section 4.1 Description

pitchdata=read.table("pitchfx.dat", header=TRUE)

halladay1=subset(pitchdata,pitchdata$pitcher=="halladay"&
    pitchdata$game==1&pitchdata$num==1)

halladay1[,c("des", "des2","pitch_type","count","new.count")]

# Section 4.2  A Sample Analysis

tim=subset(pitchdata, pitchdata$pitcher=="lincecum")

table(tim$pitch_type)

with(subset(tim, tim$new.count.type=="PA event"),
  table(brief_event))

########### figure showing movement of three type of pitches

palette(gray(seq(0,.9,len=25)))
scale.speed=function(x)
  25-25*(x-53)/(93-53)
tim.FF=subset(tim,tim$pitch_type=="FF")
tim.CH=subset(tim,tim$pitch_type=="CH")
tim.CU=subset(tim,tim$pitch_type=="CU")
with(tim.FF,plot(pfx_x,pfx_z,
  xlim=c(-13,13),ylim=c(-15,18),
  col=scale.speed(end_speed),pch=1,
  xlab="Horizontal Break",ylab="Vertical Break"))
with(tim.CU,points(pfx_x,pfx_z,
  col=scale.speed(end_speed),pch=2))
with(tim.CH,points(pfx_x,pfx_z,
  col=scale.speed(end_speed),pch=3))
text(c(-5,-12,-12),c(-7,2,15),
  c("Curveball","Changeup","Fastball"))

######## figure showing location of fastballs and curveballs

library(KernSmooth)
plot2D=function(loc.pitches,...)
{
est=bkde2D(loc.pitches,bandwidth=c(0.3,0.3))
contour(est$x1, est$x2, est$fhat,
  xlim=c(-2,2),ylim=c(0,5),
  levels=seq(.05,.3,by=.05),
  xlab="Horizontal Location",
  ylab="Vertical Location",...)
lines(c(-1,1,1,-1,-1),
  c(1.67,1.67,3.5,3.5,1.67),lwd=3)
}

tim.FF.R=subset(tim.FF,tim.FF$stand=="R")
plot2D(with(tim.FF.R,cbind(px,pz)))

tim.CU.R=subset(tim.CU,tim.CU$stand=="R")
plot2D(with(tim.CU.R,cbind(px,pz)))

######################################################