cfbfastR workflow script for one week, all plays, all teams, for LS 590 Linked Data

week_script.R

## William Blackerby
## LS 590 Linked Data
## University of Alabama SLIS

## experiment in using cfbdfastR to wrangle play-by-play data from
## http://collegefootballdata.com API, adapted to wrangle every play
## from one week of a season

# copied and pasted (with unused packages removed) from
# https://cfbfastr.sportsdataverse.org/articles/intro.html
if (!requireNamespace('pacman', quietly = TRUE)){
  install.packages('pacman')
}
pacman::p_load(tidyverse, cfbfastR, lubridate)

args <- commandArgs(trailingOnly=TRUE)
year <- strtoi(args[1])
season_type <- args[2]
week <- strtoi(args[3])


# constant: Q codes for play types, gets used for recoding during wrangling phase later
replacements <- c(
  `End Period` = "Q827",
  `Pass Incompletion` = "Q828",
  `Pass Completion` = "Q829",
  `Rush` = "Q830",
  `Pass Interception` = "Q831",
  `Sack` = "Q832",
  `Penalty` = "Q833",
  `Fumble Recovery (Own)` = "Q834",
  `Kickoff Return (Offense)` = "Q835",
  `Kickoff Return (Defense)` = "Q836",
  `Punt Return` = "Q837",
  `Two Point Pass` = "Q838",
  `Two Point Rush` = "Q839",
  `Blocked Punt` = "Q840",
  `Blocked Field Goal` = "Q841",
  `Safety` = "Q842",
  `Timeout` = "Q843",
  `Pass Reception` = "Q844",
  `Pass Interception Return` = "Q845",
  `Fumble Recovery (Opponent)` = "Q846",
  `Kickoff Return Touchdown` = "Q847",
  `Punt Return Touchdown` = "Q848",
  `Interception Return Touchdown` = "Q849",
  `Blocked Punt Touchdown` = "Q850",
  `Blocked Field Goal Touchdown` = "Q851",
  `Fumble Return Touchdown` = "Q852",
  `Missed Field Goal Return` = "Q853",
  `Missed Field Goal Return Touchdown` = "Q854",
  `Blocked PAT` = "Q855",
  `Pass` = "Q856",
  `Punt` = "Q857",
  `Kickoff` = "Q858",
  `2pt Conversion` = "Q859",
  `Defensive 2pt Conversion` = "Q860",
  `Field Goal Good` = "Q861",
  `Field Goal Missed` = "Q862",
  `Extra Point Good` = "Q863",
  `Extra Point Missed` = "Q864",
  `Interception` = "Q865",
  `End of Half` = "Q866",
  `End of Game` = "Q867",
  `Passing Touchdown` = "Q868",
  `Rushing Touchdown` = "Q869",
  `placeholder` = "Q870",
  `Offensive 1pt Safety` = "Q871",
  `Uncategorized` = "Q872"
)

# offense and defense columns for yard_line wrangling
teams <- read_csv("Teams.csv", show_col_types = FALSE) %>% select(school, abbreviation)
offense <- teams %>% select(offense_play = school, offense_abbrev = abbreviation)
defense <- teams %>% select(defense_play = school, defense_abbrev = abbreviation)

week1 <- cfbfastR::cfbd_pbp_data(year = year, season_type = season_type, week = week) %>% 

  # get only the columns we want
  select(id_play:wallclock) %>%
  
  # join helper columns for yard_line wrangling
  inner_join(offense, by = "offense_play") %>% 
  inner_join(defense, by = "defense_play") %>% 
  
  # wrangle the columns
  mutate(
    # concat clock.minutes and padded seconds into temporary column called clock
    clock = str_c(clock.minutes, str_pad(clock.seconds, 2, "left", "0"), sep = ":"),
    
    # yard_line wrangling
    yard_line = ifelse(yard_line > 50, 100 - yard_line, yard_line),
    yard_line = case_when(yards_to_goal > 50 ~ str_c(offense_abbrev, yard_line, sep = " "),
                          yards_to_goal < 50 ~ str_c(defense_abbrev, yard_line, sep = " "),
                          TRUE ~ as.character(yard_line)),
    
    # down and distance
    # down from cardinal to ordinal
    down = case_when(down == 1 ~ "1st",
                     down == 2 ~ "2nd",
                     down == 3 ~ "3rd",
                     down == 4 ~ "4th",
                     TRUE ~ as.character(down)),
    down_and_distance = str_c(down, distance, sep = " and "),
    
    # play_type to q value
    play_type = recode_factor(play_type, !!!replacements),
    
    # downs
    down = case_when(down == "1st" ~ "Q303",
                     down == "2nd" ~ "Q304",
                     down == "3rd" ~ "Q305",
                     down == "4th" ~ "Q306",
                     TRUE ~ as.character(down)),
    
    # new columns
    qid = "",
    Len = str_c(play_text, str_c(c("("), id_play, c(")")), sep = " "),
    Den = str_glue("Game action from the {year(wallclock)} {home} versus {away} Football Game"),
    P1 = "Q1585"
  ) %>% 
  
  # format strings for quick statements
  mutate(
    id_play = str_glue('""""{id_play}"'),
    yard_line = str_glue('""""{yard_line}"'),
    clock = str_glue('""""{clock}"'),
    play_text = str_glue('""""{play_text}"'),
    down_and_distance = str_glue('""""{down_and_distance}"'),
    wallclock = str_glue('""""{wallclock}"')
  ) %>% 
  
  # select columns we want and rename to corresponding P codes
  select(
    qid,
    Len,
    Den,
    P1, # "Q1585"
    P54 = id_play,
    P16 = offense_play,
    P17 = defense_play,
    P18 = game_id,
    P19 = drive_id,
    P21 = period,
    P23 = yard_line,
    P22 = clock,
    P24 = down,
    P25 = distance,
    P55 = scoring,
    P56 = yards_gained,
    P26 = play_type,
    P57 = play_text,
    P31 = down_and_distance,
    P78 = wallclock
  ) %>% 
  
  write_csv("wrangled_week.csv")

Revision 2 incorporates simple command line functionality. Needs to be built out more.