Compare batting trajectories by scraping baseball-reference data

compare_batting_trajectories.R

compare_batting_trajectories <- function(Names, 
                                         table="batting_value",
                                         stat="oWAR",
                                         NCOL=1, 
                                      playerIDs=FALSE){
  # table value - one of "batting_standard", "batting_value"
  require(Lahman)
  require(XML)
  require(ggplot2)
  require(ggthemes)
  require(dplyr)
  
  J <- length(Names)
  if(length(playerIDs==1)) playerIDs <- rep(playerIDs, J)
  
  ids <- vector(mode = "character", length = length(Names))
  for(j in 1:length(Names)){
    if (playerIDs[j]==FALSE) {
      name <- unlist(strsplit(Names[j], split=" "))
      ids[j] <- filter(Master, nameFirst==name[1], 
                   nameLast==name[2])$playerID
    } else {
      ids[j] <- Names[j]
      B <- filter(Master, playerID==ids[j])
      Names[j] <- paste(B$nameFirst, B$nameLast)
    }
  }
  
  data <- NULL
  for(j in 1:length(ids)){
     filename <- paste("http://www.baseball-reference.com/players/",
              substr(ids[j], 1, 1), "/", ids[j], ".shtml", sep="")
     d <- readHTMLTable(filename)[[table]]
     # remove minor league stats
     d <- filter(d, Lg=="AL" | Lg=="NL")
     # sum over multiple teams per season
     if(table=="batting_value"){
           d$Stat <- as.numeric(as.character(d[, stat]))
           S <- summarize(group_by(d, Age), Stat=sum(Stat))
           names(S)[2] <- stat}
     if(table=="batting_standard"){
       d <- mutate(d, Year=as.numeric(substr(Year, 1, 4)))
       mt_seasons <- filter(d, Tm=="TOT")$Year
       s_seasons <- d$Year[!d$Year %in% mt_seasons]
       S <- filter(d, (Year %in% s_seasons) |
                      (Year %in% mt_seasons & Tm=="TOT"))
       S[, stat] <- as.numeric(as.character(S[, stat]))
     }
     d_player <- data.frame(Player=Names[j],
                      Age=as.numeric(as.character(S$Age)),
            Statistic=S[, stat])
     data <- rbind(data, d_player)
  }
  names(data)[3] <- "Statistic"
  print(ggplot(data, aes(Age, Statistic)) + 
          geom_point(color="red") + geom_smooth() +
          facet_wrap(~ Player, ncol=NCOL) +
          theme_fivethirtyeight() +
          theme(plot.title = element_text(size = rel(2), hjust=0.5,
                                          color = "blue")) +
          theme(strip.text = element_text(size = rel(1.5), hjust=0.5,
                                          color = "red")) +
          ggtitle(paste(stat, "Career Trajectories")))
  
  data
}