MVPTrack.R

# constants
YEAR_START<-2000
YEAR_END<-2017

#######################################################
#################### MVP DATA ######################### 
#######################################################
# GRAB MVP DATA FROM EACH SEASON
mvp_dat<-data.frame(Rank=integer(),Player=character(),Age=double(),Tm=character(),
                    First=double(),Pts.Won=double(),Pts.Max=double(),Share=double(),
                    G=double(),MP=double(),PTS=double(),TRB=double(),AST=double(),
                    STL=double(),BLK=double(),FG.=double(),X3P.=double(),FT.=double(),
                    WS=double(),WS.48=double(),Season=double())
for (year in YEAR_START:YEAR_END) {
  str<-paste("C:/Users/Caio/repos/nba-models/award-stats/",year,".csv",sep="")
  dat_temp<-read.csv(str, header = TRUE,stringsAsFactors=FALSE)
  dat_temp$Season<-year
  # normalize stats
  ## G
  dat_temp$G<-dat_temp$G/max(dat_temp$G)
  ## MP
  dat_temp$MP<-dat_temp$MP/max(dat_temp$MP)
  ## PTS
  dat_temp$PTS<-dat_temp$PTS/max(dat_temp$PTS)
  ## TRB
  dat_temp$TRB<-dat_temp$TRB/max(dat_temp$TRB)
  ## AST
  dat_temp$AST<-dat_temp$AST/max(dat_temp$AST)
  ## STL
  dat_temp$STL<-dat_temp$STL/max(dat_temp$STL)
  ## BLK
  dat_temp$BLK<-dat_temp$BLK/max(dat_temp$BLK)
  ## FG.
  dat_temp$FG.<-dat_temp$FG./max(dat_temp$FG.)
  ## FT.
  dat_temp$FT.<-dat_temp$FT./max(dat_temp$FT.)
  ## WS.48
  dat_temp$WS.48<-dat_temp$WS.48/max(dat_temp$WS.48)
  
  # clean player names
  dat_temp$Player<-as.character(dat_temp$Player)
  names<-strsplit(as.character(dat_temp$Player),"[\\\\]")
  for (idx in 1:dim(dat_temp)[1]) {
    dat_temp$Player[idx]<-names[[idx]][1]
  }
  
  mvp_dat<-data.frame(rbind.data.frame(as.matrix(mvp_dat), as.matrix(dat_temp)))
}
head(mvp_dat)

# set each column to appropriate data type
levels(mvp_dat$Rank)<-c(as.numeric(levels(mvp_dat$Rank)))
mvp_dat$Age<-type.convert(mvp_dat$Age)
#class(as.numeric(levels(mvp_dat$Age))[mvp_dat$Age])
mvp_dat$Pts.Won<-type.convert(mvp_dat$Pts.Won)
mvp_dat$First<-type.convert(mvp_dat$First)
mvp_dat$G<-type.convert(mvp_dat$G)
mvp_dat$MP<-type.convert(mvp_dat$MP)
mvp_dat$PTS<-type.convert(mvp_dat$PTS)
mvp_dat$TRB<-type.convert(mvp_dat$TRB)
mvp_dat$AST<-type.convert(mvp_dat$AST)
mvp_dat$STL<-type.convert(mvp_dat$STL)
mvp_dat$BLK<-type.convert(mvp_dat$BLK)
mvp_dat$FG.<-type.convert(mvp_dat$FG.)
mvp_dat$FT.<-type.convert(mvp_dat$FT.)
mvp_dat$WS.48<-type.convert(mvp_dat$WS.48)
mvp_dat$Season<-type.convert(mvp_dat$Season)

# full mod
mvp.mod<-lm(First~Age+G+MP+PTS+TRB+AST+STL+BLK+FG.+WS.48,data=mvp_dat)
summary(mvp.mod)

#reduced mod
mvp.mod.red<-lm(First~G+PTS+AST+WS.48,data=mvp_dat)
summary(mvp.mod.red)

# make predictions
pred<-fitted(mvp.mod)
mvps<-data.frame(Rank=integer(),Player=character(),Age=double(),Tm=character(),
                 First=double(),Pts.Won=double(),Pts.Max=double(),Share=double(),
                 G=double(),MP=double(),PTS=double(),TRB=double(),AST=double(),
                 STL=double(),BLK=double(),FG.=double(),X3P.=double(),FT.=double(),
                 WS=double(),WS.48=double(),Season=double())
for (year in levels(as.factor(mvp_dat$Season))) {
  dat_temp<-mvp_dat[which(mvp_dat$Season==year),]
  temp_preds<-pred[which(mvp_dat$Season==year)]
  mvp<-dat_temp[which(temp_preds==max(temp_preds)),]
  mvps<-data.frame(rbind(as.matrix(mvps), as.matrix(mvp)))
}


# calculate accuracy
## hack to fix seasons like 1995,2009,2016 where rank is messed up
#levels(mvps$Rank)<-c(as.numeric(levels(mvps$Rank)))
truevals<-mvp_dat[which(mvp_dat$Rank==1),]
errs<-mvps[which(mvps$Rank!=1),]
accuracy<-1-(dim(errs)[1]/dim(mvps)[1])
accuracy

#######################################################
################   SEASON STANDINGS    ################ 
#######################################################
# GRAB DATA FROM EACH SEASON
dat_std<-data.frame(Rk=integer(),Team=character(),Overall=character(),Home=character(),Road=character(),
                    Pre=character(),Post=character(),X3=character(),X10=character(),Oct=character(),
                    Nov=character(),Dec=character(),Jan=character(),Feb=character(),Mar=character(),
                    Apr=character(),Season=factor())
for (year in YEAR_START:YEAR_END) {
  # read data
  str<-paste("C:/Users/Caio/repos/nba-models/season-standings/",year,".csv",sep="")
  dat_temp<-read.csv(str, header = TRUE)
  
  # add season column
  dat_temp$Season<-year
  
  # fix missing Oct data
  if(!("Oct" %in% colnames(dat_temp)))
  {
    dat_temp$Oct<-NA
  }
  if(!("Nov" %in% colnames(dat_temp)))
  {
    dat_temp$Nov<-NA
  }
  
  # fix columns
  dat_temp<-data.frame(Rk=dat_temp$Rk,Team=dat_temp$Team,Overall=dat_temp$Overall,Home=dat_temp$Home,
                      Road=dat_temp$Road,Pre=dat_temp$Pre,Post=dat_temp$Post,X3=dat_temp$X3,X10=dat_temp$X10,
                      Oct=dat_temp$Oct,Nov=dat_temp$Nov,Dec=dat_temp$Dec,Jan=dat_temp$Jan,Feb=dat_temp$Feb,
                      Mar=dat_temp$Mar,Apr=dat_temp$Apr,Season=dat_temp$Season)
  
  # add season to main dataframe
  dat_std<-data.frame(rbind(as.matrix(dat_std), as.matrix(dat_temp)))
}
head(dat_std)

# add wins and losses
wl<-strsplit(as.character(dat_std$Overall), "-")
dat_std$Wins<-0
dat_std$Losses<-0
for (idx in 1:dim(dat_std)[1]) {
  dat_std[idx,]$Wins<-wl[[idx]][1]
  dat_std[idx,]$Losses<-wl[[idx]][2]
}

#######################################################
################ SEASON PLAYER TOTALS ################
#######################################################
# GRAB DATA FROM EACH SEASON
dat_totals<-data.frame(Rk=integer(),Player=character(),Pos=character(),Age=double(),Tm=character(),
                    G=double(),GS=double(),MP=double(),FG=double(),FGA=double(),FG.=double(),
                    X3P=double(),X3PA=double(),X3P.=double(),X2P=double(),X2PA=double(),
                    X2P.=double(),eFG.=double(),FT=double(),FTA=double(),FT.=double(),ORB=double(),
                    DRB=double(),TRB=double(),AST=double(),STL=double(),BLK=double(),TOV=double(),
                    PF=double(),PTS=double(),Season=integer())
for (year in YEAR_START:YEAR_END) {
  # read data
  str<-paste("C:/Users/Caio/repos/nba-models/season-stats-totals/",year,".csv",sep="")
  dat_temp<-read.csv(str, header = TRUE)
  
  # add season column
  dat_temp$Season<-year
  
  # clean player names
  dat_temp$Player<-as.character(dat_temp$Player)
  names<-strsplit(as.character(dat_temp$Player),"[\\\\]")
  for (idx in 1:dim(dat_temp)[1]) {
    name<-names[[idx]][1]
    name<-gsub("[*]","",name)
    dat_temp$Player[idx]<-name
  }
  
  # might consider lockout seasons
  
  # add season to main dataframe
  dat_totals<-data.frame(rbind(as.matrix(dat_totals), as.matrix(dat_temp)))
}
head(dat_totals)

## add MVP winning seasons
dat_totals$MVP<-FALSE
for (idx in 1:dim(truevals)[1]) {
  dat_totals[which(as.character(dat_totals$Player)==as.character(truevals[idx,]$Player)&dat_totals$Season==truevals[idx,]$Season),]$MVP<-TRUE
}

# add first place votes
dat_totals$First<-0
for (idx in 1:dim(mvp_dat)[1]) {
  dat_totals[which(as.character(dat_totals$Player)==as.character(mvp_dat[idx,]$Player)&dat_totals$Season==mvp_dat[idx,]$Season),]$First<-mvp_dat[idx,]$First
}

#######################################################
################ SEASON PLAYER PER GAME################ 
#######################################################
# GRAB DATA FROM EACH SEASON
dat_pg<-data.frame(Rk=integer(),Player=character(),Pos=character(),Age=double(),Tm=character(),
                     G=double(),GS=double(),MP=double(),FG=double(),FGA=double(),FG.=double(),
                     X3P=double(),X3PA=double(),X3P.=double(),X2P=double(),X2PA=double(),
                     X2P.=double(),eFG.=double(),FT=double(),FTA=double(),FT.=double(),ORB=double(),
                     DRB=double(),TRB=double(),AST=double(),STL=double(),BLK=double(),TOV=double(),
                     PF=double(),PS.G=double(),Season=integer())
for (year in YEAR_START:YEAR_END) {
  # read data
  str<-paste("C:/Users/Caio/repos/nba-models/season-stats-pergame/",year,".csv",sep="")
  dat_temp<-read.csv(str, header = TRUE)
  
  # add season column
  dat_temp$Season<-year
  
  # clean player names
  dat_temp$Player<-as.character(dat_temp$Player)
  names<-strsplit(as.character(dat_temp$Player),"[\\\\]")
  for (idx in 1:dim(dat_temp)[1]) {
    name<-names[[idx]][1]
    name<-gsub("[*]","",name)
    dat_temp$Player[idx]<-name
  }
  
  # might consider lockout seasons
  
  # add season to main dataframe
  dat_pg<-data.frame(rbind(as.matrix(dat_pg), as.matrix(dat_temp)))
}
head(dat_pg)

## add MVP winning seasons
dat_pg$MVP<-FALSE
for (idx in 1:dim(truevals)[1]) {
  dat_pg[which(as.character(dat_pg$Player)==as.character(truevals[idx,]$Player)&dat_pg$Season
               ==truevals[idx,]$Season),]$MVP<-TRUE
}

# add first place votes
dat_pg$First<-0
for (idx in 1:dim(mvp_dat)[1]) {
  dat_pg[which(as.character(dat_pg$Player)==as.character(mvp_dat[idx,]$Player)&
                 dat_pg$Season==mvp_dat[idx,]$Season),]$First<-mvp_dat[idx,]$First
  
}

# test truevals
dat_pg[which(dat_pg$MVP==TRUE),]

# fix data.frame classes
for (idx in c(4,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30)) {
  class(dat_pg[,idx])
  dat_pg[,idx]<-as.numeric(type.convert(dat_pg[,idx]))
  class(dat_pg[,idx])
}


# clean missing observations
dat_pg_clean<-dat_pg[complete.cases(dat_pg), ]

#subset data
dat.mod<-dat_pg_clean[which(dat_pg_clean$G>60),]
dat.mod<-dat.mod[which(dat.mod$Tm!="TOT"),]

# normalize data
for (year in levels(as.factor(dat.mod$Season))) {
  for (idx in c(4,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30)) {
    dat.mod[,idx]<-dat.mod[,idx]/max(dat.mod[,idx])
  }
}

# fix team strings
## standings full team name, player stats abbreviated
abbrev<-c("ATL","BOS","BRK","CHA","CHH","CHI","CHO","CLE","DAL","DEN","DET","GSW","HOU","IND","LAC","LAL",
          "MEM","MIA","MIL","MIN","NJN","NOH","NOK","NOP","NYK","OKC","ORL","PHI","PHO","POR","SAC","SAS",
          "SEA","TOR","UTA","VAN","WAS")
fullnames<-c("Atlanta Hawks","Boston Celtics","Brooklyn Nets","Charlotte Bobcats",
  "Charlotte Hornets","Chicago Bulls","Charlotte Hornets","Cleveland Cavaliers","Dallas Mavericks","Denver Nuggets",
  "Detroit Pistons","Golden State Warriors","Houston Rockets","Indiana Pacers","Los Angeles Clippers",
  "Los Angeles Lakers","Memphis Grizzlies","Miami Heat","Milwaukee Bucks","Minnesota Timberwolves",
  "New Jersey Nets","New Orleans Hornets","New Orleans/Oklahoma City Hornets","New Orleans Pelicans",
  "New York Knicks","Oklahoma City Thunder","Orlando Magic","Philadelphia 76ers","Phoenix Suns",
  "Portland Trail Blazers","Sacramento Kings","San Antonio Spurs",
  "Seattle SuperSonics","Toronto Raptors","Utah Jazz","Vancouver Grizzlies","Washington Wizards")
names(fullnames)<-abbrev

# add team wins
dat.mod$Team.Wins<-0
for (team_abbrev in levels(as.factor(dat.mod$Tm))) {
  for (season in levels(as.factor(dat.mod$Season))) {
    team_name<-fullnames[team_abbrev]
    roster<-dat.mod[which(as.character(dat.mod$Tm)==team_abbrev & as.character(dat.mod$Season)==season),]
    if (dim(roster)[1]!=0) {
      roster$Team.Wins<-dat_std[which(as.character(dat_std$Team)==team_name & as.character(dat_std$Season)==season),]$Wins
    }
    dat.mod[which(as.character(dat.mod$Tm)==team_abbrev & as.character(dat.mod$Season)==season),]<-roster
  }
}
dat.mod$Team.Wins<-type.convert(dat.mod$Team.Wins)

# baseline model
mod<-lm(First~Pos+Age+G+GS+MP+FG+FGA+FG.+X3P+X3PA+X3P+X2P+X2PA+X2P.+eFG.+FT+FTA+FT.+ORB+DRB+TRB+AST+STL+BLK+TOV+PF+PS.G+Team.Wins,data=dat.mod)
summary(mod)

# reduced model
mod.red<-lm(First~Pos+G+MP+eFG.+ORB+DRB+TRB+AST+STL+BLK+TOV+PF+PS.G+Team.Wins,data=dat.mod)
summary(mod.red)

# make predictions
pred<-fitted(mod.red)
mvps<-data.frame(Rk=integer(),Player=character(),Pos=character(),Age=double(),Tm=character(),
                 G=double(),GS=double(),MP=double(),FG=double(),FGA=double(),FG.=double(),
                 X3P=double(),X3PA=double(),X3P.=double(),X2P=double(),X2PA=double(),
                 X2P.=double(),eFG.=double(),FT=double(),FTA=double(),FT.=double(),ORB=double(),
                 DRB=double(),TRB=double(),AST=double(),STL=double(),BLK=double(),TOV=double(),
                 PF=double(),PS.G=double(),Season=integer(),MVP=logical(),First=double(),Team.Wins=integer())
for (year in levels(as.factor(dat.mod$Season))) {
  dat_temp<-dat.mod[which(dat.mod$Season==year),]
  temp_preds<-pred[which(dat.mod$Season==year)]
  mvp<-dat_temp[which(temp_preds==max(temp_preds)),]
  mvps<-data.frame(rbind(as.matrix(mvps), as.matrix(mvp)))
}
mvps

# calculate accuracy
truevals<-dat_pg_clean[which(dat_pg_clean$MVP==TRUE),]
errs<-mvps[which(mvps$MVP==FALSE),]
accuracy<-1-(dim(errs)[1]/dim(mvps)[1])
accuracy