R work for why are baseball games so long post

baseball_durations.R

library(tidyverse)
library(CalledStrike)
library(broom)

######################################
# function retro_work()
######################################
# given a Retrosheet play-by-play dataset 
# collects the number of PAs and 
# number of pitches for all games in 
# specific season

retro_work <- function(d){
  d %>% 
    mutate(pseq = gsub("[.>123N+*]", "", PITCH_SEQ_TX),
         N.Pitches = nchar(pseq)) -> d

  d %>% 
    group_by(GAME_ID) %>% 
    summarize(PA = n(),
              N_Pitches = sum(N.Pitches),
            maxINNING = max(INN_CT)) %>% 
    filter(maxINNING == 9) -> S
  S$GAME_ID <- as.character(S$GAME_ID)
  S
}

########################################
# function one_season()
########################################
# inputs are the season number and the output from
# the retro_work() function
# function will read in the Retrosheet game logs 
# dataset from disk and merge the game duration variable
# with the retroS data frame

one_season <- function(season, retroS){
  filename <- paste("~/Dropbox/Google Drive/gamelogs/gamelogs/gl",
                    season, ".txt", sep = '')
  gdata <- read_csv(filename, col_names = FALSE)
  header <- read_csv("~/Dropbox/Google Drive/gamelogs/gamelogs/game_log_header.csv")
  names(gdata) <- names(header)
  gdata %>% 
    mutate(GAME_ID = paste(HomeTeam, Date, DoubleHeader, 
                           sep = "")) -> gdata
  inner_join(retroS, select(gdata, GAME_ID, Duration),
                  by = "GAME_ID") -> retroS
  retroS$Season <- season
  retroS
}

# using the two functions

# load the Retrosheet play-by-play dataset on disk
load("~/Dropbox/Google Drive/Retrosheet/pbp.2019.Rdata")

S2019 <- retro_work(d2019)
S2019 <- one_season(2019, S2019)

# I repeated this operation for all seasons 2000 through
# 2019 -- row merged these data frames into the data frame
# Sall

# graph of the mean duration of the games against season

Sall %>% 
  group_by(Season) %>% 
  summarize(M = mean(Duration)) %>% 
  ggplot(aes(Season, M)) +
  geom_point(size = 4, color = "red") +
  geom_smooth(method = "loess", se = FALSE) +
  increasefont() +
  ylab("Mean Duration (Min)") +
  ggtitle("Mean Length of a 9 Inning Game") +
  centertitle()

# graph of the mean number of PAs against season

Sall %>% 
  group_by(Season) %>% 
  summarize(M = mean(PA)) %>% 
  ggplot(aes(Season, M)) +
  geom_point(size = 4, color = "red") +
  geom_smooth(method = "loess", se = FALSE) +
  increasefont() +
  ylab("Mean PA") +
  ggtitle("Mean Number of PA of a 9 Inning Game") +
  centertitle()

# graph of the mean number of pitches per PA

Sall %>% 
  group_by(Season) %>% 
  summarize(M = mean(N_Pitches / PA)) %>% 
  ggplot(aes(Season, M)) +
  geom_point(size = 4, color = "red") +
  geom_smooth(method = "loess", se = FALSE) +
  increasefont() +
  ylab("Mean Pitches / PA") +
  ggtitle("Mean Number of Pitches per PA") +
  centertitle()

# graph of the mean number of pitches per game

Sall %>% 
  group_by(Season) %>% 
  summarize(M = mean(N_Pitches)) %>% 
  ggplot(aes(Season, M)) +
  geom_point(size = 4, color = "red") +
  geom_smooth(method = "loess", se = FALSE) +
  increasefont() +
  ylab("Mean Pitches per Game") +
  ggtitle("Mean Number of Pitches per Game") +
  centertitle()

# fit regressions of (N_Pitches, Duration) for
# all seasons

regressions <- Sall %>% group_by(Season) %>%
  do(tidy(lm(Duration ~ N_Pitches, data=.)))

# graph of regression slopes against season

regressions %>% 
  filter(term == "N_Pitches") %>% 
  ggplot(aes(Season, estimate)) +
  geom_point(size = 4, color = "red") +
  geom_smooth(method = "loess", se = FALSE) +
  increasefont() +
  ylab("Mean Time Per Pitch (Min)") +
  ggtitle("Mean Time Per Pitch") +
  centertitle()